Industrial park-oriented multi-pollution source collaborative treatment and emission optimization control method

Abstract

The application provides an industrial park-oriented multi-pollution source collaborative treatment and emission optimization control method, relates to the technical field of environmental pollution treatment, and comprises the following steps: obtaining real-time monitoring data and meteorological data, generating emission characteristic fingerprint data and pollutant distribution weight coefficients, generating pollutant transmission path data by using a two-way fluid-solid coupling calculation method and dividing a control partition, constructing a pollutant migration and diffusion model to calculate a pollution source correlation strength matrix, determining optimal emission reduction proportions and dynamic emission control parameters, and establishing a multi-pollution source linkage control strategy. The application realizes accurate identification and collaborative treatment of multi-pollution sources and improves pollution treatment efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to environmental pollution control technology, and in particular to a multi-pollution source collaborative control and emission optimization regulation method for industrial parks. BACKGROUND

[0002] With the continuous expansion of industrial parks, the interaction between multiple pollution sources is increasingly complex, and traditional single pollution source control methods cannot meet the environmental control requirements. Existing technologies mainly control based on fixed emission standards, lack in-depth analysis of the correlation between pollution sources, and are difficult to achieve precise control.

[0003] Currently, there are three main problems in the collaborative control of pollution sources in industrial parks: first, it is difficult to accurately identify the emission characteristics of each pollution source and their mutual influence; second, the complex processes of gas-liquid phase transformation, chemical reaction and sedimentation of pollutants during transmission are not fully considered; third, there is a lack of differentiated regulation mechanism based on the correlation strength of pollution sources.

[0004] Therefore, it is urgent to establish a multi-pollution source collaborative control and emission optimization regulation method for industrial parks, which analyzes the emission characteristic fingerprints and transmission paths of pollution sources, constructs a pollutant migration and diffusion model, and realizes differentiated regulation based on the correlation strength of pollution sources, thereby improving the scientificity and accuracy of environmental governance in industrial parks. SUMMARY

[0005] The embodiments of the present application provide a multi-pollution source collaborative control and emission optimization regulation method for industrial parks, which can solve the problems in the prior art.

[0006] The first aspect of the embodiments of the present application is,

[0007] A multi-pollution source collaborative control and emission optimization regulation method for industrial parks is provided, comprising:

[0008] Obtaining real-time monitoring data and meteorological data of multiple pollution sources in an industrial park;

[0009] Based on the real-time monitoring data, the emission characteristic fingerprint data of each pollution source is generated by analyzing the characteristic pollutant component concentration and the ratio relationship of the characteristic pollutant component concentration of each pollution source, and the pollutant distribution weight coefficient is calculated according to the emission characteristic fingerprint data;

[0010] According to the emission characteristic fingerprint data, the meteorological data and the pollutant distribution weight coefficient, the gas-liquid phase transformation rate, the chemical reaction rate and the sedimentation rate of the pollutants are taken as calculation parameters by using a two-way fluid-structure coupling calculation method to generate pollutant transmission path data, and the control partition of the industrial park is divided based on the pollutant transmission path data;

[0011] According to the control partition and the pollutant transmission path data, a pollutant migration and diffusion model is constructed, a pollutant superposition influence coefficient between each pollution source is calculated, a pollution source correlation strength matrix is generated, an optimal emission reduction proportion of each pollution source is calculated based on the pollution source correlation strength matrix and a pollutant distribution weight coefficient, dynamic emission control parameters of each pollution source are calculated according to the optimal emission reduction proportion, a multi-pollution source linkage control strategy is established based on the dynamic emission control parameters and the pollutant transmission path data, and collaborative control instructions of each pollution source are generated.

[0012] In an alternative embodiment,

[0013] Based on real-time monitoring data, by analyzing the characteristic pollutant component concentration of each pollution source and the ratio of the characteristic pollutant component concentration, the emission characteristic fingerprint data of each pollution source is generated, and the pollutant distribution weight coefficient is calculated according to the emission characteristic fingerprint data, including:

[0014] Obtain real-time monitoring data of multiple pollution sources in the industrial park, the real-time monitoring data includes characteristic pollutant component concentration data, perform wavelet transform decomposition on the characteristic pollutant component concentration data to obtain first frequency band component concentration data and second frequency band component concentration data;

[0015] Based on the first frequency band component concentration data and the second frequency band component concentration data, a chemical reaction kinetics correction model is constructed, the direct reaction amount of the pollutant, the synergistic reaction amount of the pollutant and the attenuation amount of the pollutant are calculated by the chemical reaction kinetics correction model, and the corrected component concentration data is generated;

[0016] The attention mechanism is used for time series analysis on the corrected component concentration data, the time distribution characteristics of the component concentration are extracted, the time characteristic data is generated, the conversion relationship between the pollutant components is calculated according to the time characteristic data, and the component conversion characteristic data is generated;

[0017] According to the time characteristic data and the component conversion characteristic data, the correlation between the pollutant components is calculated, the component action coefficient is generated, the time characteristic data and the component conversion characteristic data are weighted and fused by using the component action coefficient, and the pollution source emission characteristic fingerprint data is obtained;

[0018] The pollution source emission characteristic fingerprint data is substituted into the optimization equation, and the pollutant distribution weight coefficient is solved under the conditions of satisfying the component action constraint and the component conversion constraint, the pollutant distribution weight coefficient is used to represent the pollutant emission contribution of each pollution source.

[0019] In an alternative embodiment,

[0020] constructing a chemical reaction kinetics correction model based on the first frequency band group component concentration data and the second frequency band group component concentration data, calculating a pollutant direct reaction amount, a pollutant synergistic reaction amount and a pollutant decay amount through the chemical reaction kinetics correction model, and generating corrected component concentration data including:

[0021] normalizing the first frequency band group component concentration data and the second frequency band group component concentration data to obtain standardized concentration data, constructing a reaction network structure based on the standardized concentration data, establishing a material conversion relationship between components as nodes, and calculating a concentration gradient between components to obtain a reaction intensity;

[0022] establishing a reaction rate equation of the chemical reaction kinetics correction model according to the reaction intensity, wherein the reaction rate equation includes a reactant concentration term, a reaction rate constant term and a correction coefficient term;

[0023] solving the reaction rate equation by using the standardized concentration data to obtain a frequency band coupling coefficient, establishing a component concentration change matrix according to the frequency band coupling coefficient, substituting the component concentration change matrix into the reaction rate equation to obtain a reaction rate value, and calculating a pollutant direct reaction amount according to a product of the reaction rate value and a time step;

[0024] constructing a component interaction matrix based on the component concentration change matrix, substituting the component interaction matrix into the reaction rate equation to calculate a pollutant synergistic reaction amount;

[0025] obtaining environmental monitoring data, correcting the correction coefficient term in the reaction rate equation according to the environmental monitoring data, and calculating a pollutant decay amount;

[0026] adding the pollutant direct reaction amount and the pollutant synergistic reaction amount to the standardized concentration data, and subtracting the pollutant decay amount to generate corrected component concentration data.

[0027] In an optional embodiment,

[0028] According to the emission characteristic fingerprint data, meteorological data and pollutant distribution weight coefficient, the gas-liquid phase state conversion rate, chemical reaction rate and settling rate of the pollutant are taken as calculation parameters by using a two-way fluid-structure coupling calculation method to generate pollutant transport path data, and the pollutant transport path data is used to divide a control partition of the industrial park including:

[0029] analyzing the concentration distribution characteristics of different pollutant components in the emission characteristic fingerprint data to calculate mutual promotion intensity and mutual inhibition intensity between pollutant components, and constructing a component interaction strength matrix;

[0030] The inter-component interaction strength matrix is combined with a pollutant distribution weight coefficient, a local pollutant concentration distribution characteristic value is calculated according to temperature and humidity information in meteorological data, a local concentration correction coefficient is obtained, a mass transfer resistance of a gas-liquid interface is corrected based on the local concentration correction coefficient, and a phase change rate considering the interface resistance is calculated by combining the inter-component interaction strength matrix;

[0031] A mass transfer flux of the gas-liquid interface is calculated according to the phase change rate, a chemical reaction rate considering local distribution characteristics is calculated by combining the mass transfer flux with the local concentration correction coefficient, and a pollutant deposition rate considering interface mass transfer characteristics is calculated based on the chemical reaction rate and the phase change rate;

[0032] The phase change rate, the chemical reaction rate and the deposition rate are substituted into a two-way fluid-solid coupling control equation, which contains an interaction term of gas phase pollutants and particulate matters, and an adaptive parameter adjustment mechanism is established according to real-time changes of environmental monitoring data, and a pollutant transport flux at each grid point is calculated by iterative solution;

[0033] The spatiotemporal variation characteristics of the transport flux are analyzed, main transport channels are identified, and a pollution load index of each transport channel is calculated to generate pollutant transport path data;

[0034] The diffusion risk coefficient of each grid point is calculated according to the pollutant transport path data, and the industrial park is divided into control zones based on the pollution load index and the diffusion risk coefficient.

[0035] In an alternative embodiment,

[0036] The spatiotemporal variation characteristics of the transport flux are analyzed, main transport channels are identified, and a pollution load index of each transport channel is calculated to generate pollutant transport path data, which includes:

[0037] The pollutant concentration spatial gradient of the transport flux is calculated by using a sliding time window method, the length of the sliding time window is dynamically adjusted according to the pollutant concentration change rate, the time sequence characteristics of the spatial gradient are analyzed, the gradient discrimination threshold is set according to the physical and chemical properties of the pollutant, the data sampling points are increased when the spatial gradient is greater than the gradient discrimination threshold, and the transport flux sampling data are obtained;

[0038] The transport flux sampling data are spatiotemporally decomposed to obtain the time evolution characteristics and the spatial distribution characteristics of the transport flux, and the spatiotemporal variation characteristics are correlated with the temperature field, the humidity field and the wind field for multi-factor correlation analysis, and a quantitative corresponding relationship between the spatiotemporal variation characteristics of the transport flux and the environmental field is established;

[0039] An environmental stress degree index is constructed based on the quantitative correspondence, the environmental stress degree index characterizes the comprehensive influence of the environmental field on the diffusion, sedimentation and accumulation of the pollutants, the environmental stress degree index and the transport flux are combined by weighting, and a pollution load index of each transport channel is calculated;

[0040] An intensity-direction phase diagram of the transport flux is constructed, the intensity contour and direction field of the phase diagram are extracted, the main transport channel is determined according to the continuity of the contour and direction field, and the spatial position, transport flux and pollution load of the main transport channel are integrated to generate pollution transport path data.

[0041] In an optional embodiment,

[0042] A pollution migration and diffusion model is constructed based on the control partition and the pollution transport path data, a pollution superposition influence coefficient between pollution sources is calculated, and a pollution source correlation strength matrix is generated, including:

[0043] The control partition is divided into grids according to the pollution transport path data, the grid division density is adaptively adjusted based on the concentration gradient calculated based on the pollution concentration monitoring data, the mass conservation equation and momentum conservation equation containing the chemical reaction rate of the pollutants, the gas-liquid phase conversion rate and the sedimentation rate are established in the grid, and the pollution migration and diffusion model is constructed;

[0044] The pollution concentration field distribution data is obtained by solving the pollution migration and diffusion model, the sensitivity coefficient representing the influence of the change of the pollution emission amount on the concentration of the receptor point is calculated based on the pollution concentration field distribution data, the sensitivity coefficient is substituted into the nonlinear response function containing the gas phase oxidation, the liquid phase oxidation and the heterogeneous reaction, and the pollution superposition influence coefficient is calculated;

[0045] The single-source transport influence coefficient is calculated based on the pollution superposition influence coefficient, the single-source transport influence coefficient is input into the transfer closure algorithm for calculating the multi-source coupling transport effect of the pollutants, the coupling transport influence coefficient is obtained, and the single-source transport influence coefficient and the coupling transport influence coefficient are combined based on the time series sliding window to generate the pollution source correlation strength matrix.

[0046] In an optional embodiment,

[0047] The optimal emission reduction proportion of each pollution source is calculated based on the pollution source correlation strength matrix and the pollution distribution weight coefficient, the dynamic emission regulation and control parameter of each pollution source is calculated according to the optimal emission reduction proportion, the multi-pollution source linkage control strategy is established based on the dynamic emission regulation and control parameter and the pollution transport path data, and the collaborative regulation and control instruction of each pollution source is generated, including:

[0048] The pollution source correlation strength matrix and the pollutant distribution weight coefficient are input into a nonlinear programming model, the nonlinear programming model includes an objective function of maximizing pollutant reduction and minimizing economic cost, and the optimal emission reduction proportion of each pollution source is obtained by solving the nonlinear programming model through a genetic algorithm;

[0049] A state space model including pollution source characteristics, meteorological conditions and environmental capacity is established according to the optimal emission reduction proportion, a Kalman filtering algorithm is used to estimate the system state in real time, and the state estimation result is input into a model predictive controller to calculate dynamic emission regulation parameters of each pollution source;

[0050] A pollution source weighted directed graph is constructed based on the dynamic emission regulation parameters and pollutant transmission path data, a community discovery algorithm is used to identify strongly correlated pollution source groups in the weighted directed graph, and a differentiated group control strategy is generated according to the dynamic emission regulation parameters of each group to establish a multi-pollution source linkage control strategy;

[0051] A hierarchical control structure is constructed based on the multi-pollution source linkage control strategy, a feedback compensation mechanism is set in the hierarchical control structure, actual pollutant concentration change data is input into the feedback compensation mechanism for dynamic correction, and collaborative regulation instructions of each pollution source are generated.

[0052] In an optional embodiment,

[0053] A pollution source weighted directed graph is constructed based on the dynamic emission regulation parameters and pollutant transmission path data, a community discovery algorithm is used to identify strongly correlated pollution source groups in the weighted directed graph, and a differentiated group control strategy is generated according to the dynamic emission regulation parameters of each group to establish a multi-pollution source linkage control strategy, including:

[0054] The dynamic emission regulation parameters are taken as node characteristic values, node connection relationships are constructed based on pollutant transmission path data, edge characteristic values are obtained by updating the transmission amount between nodes in an exponential decay manner, and a pollution source weighted directed graph is constructed according to the node characteristic values and the edge characteristic values;

[0055] Actual transmission strength and network reference transmission strength between nodes are calculated based on the pollution source weighted directed graph, the actual transmission strength and the network reference transmission strength are input into a community discovery algorithm, and an initial group division scheme is obtained by iteratively optimizing group division indicators;

[0056] The inter-group transmission distance coefficient and the meteorological similarity coefficient are calculated based on the initial group division scheme, the transmission distance coefficient and the meteorological similarity coefficient are combined with a group division index to construct a group optimization criterion, and the initial group division scheme is optimized according to the group optimization criterion to obtain a target pollution source group; for the target pollution source group, a group control model including a pollutant reduction amount and a control cost is constructed, and a pollution source group control target is obtained by solving the group control model;

[0057] The connection degree coefficient, the transmission flux coefficient and the position correlation coefficient of each node in the pollution source group are calculated, the connection degree coefficient, the transmission flux coefficient and the position correlation coefficient are combined to obtain a node priority coefficient, and the nodes in the pollution source group are graded based on the node priority coefficient to obtain a group internal grading scheme.

[0058] According to the group internal grading scheme and the pollution source group control target, a group game model including a pollutant reduction amount and a control cost is constructed, a differentiated group control strategy is obtained by solving the group game model, a real-time monitoring data is collected to calculate a control correction amount, the control correction amount is input into a compensation model to adjust the differentiated group control strategy in real time, and a multi-pollution source linkage control strategy is established.

[0059] The second aspect of the embodiment of the application,

[0060] An electronic device is provided, comprising:

[0061] A processor;

[0062] A memory for storing processor-executable instructions;

[0063] The processor is configured to invoke the instructions stored in the memory to execute the method described above.

[0064] The third aspect of the embodiment of the application,

[0065] A computer-readable storage medium is provided, which stores computer program instructions, and the computer program instructions are executed by a processor to implement the method described above.

[0066] In the embodiment, by acquiring real-time monitoring data and meteorological data of multiple pollution sources in the industrial park, emission characteristic fingerprint data and pollutant distribution weight coefficients are generated, precise identification and quantitative analysis of multiple pollution sources are realized, and the accuracy and timeliness of pollution source analysis are improved. A more accurate pollutant migration and diffusion model is constructed by using a two-way fluid-structure coupling calculation method, taking the gas-liquid phase conversion rate, chemical reaction rate and settling rate of the pollutant as calculation parameters, which can accurately reflect the mutual influence relationship between different pollution sources and provide a scientific basis for collaborative governance. The optimal emission reduction proportion is calculated based on the pollution source correlation strength matrix and the pollutant distribution weight coefficient, dynamic emission control parameters and multi-pollution source linkage control strategy are generated, and collaborative governance and emission optimization and control of multiple pollution sources in the industrial park are realized, which not only ensures that the environmental quality meets the standard, but also maximizes the reduction of enterprise emission reduction cost, and improves the scientificity and economy of environmental management. BRIEF DESCRIPTION OF DRAWINGS

[0067] Figure 1 A flowchart of the method for collaborative governance and emission optimization and control of multiple pollution sources in the industrial park according to the embodiment of the present application is shown in

[0068] Figure 2 A pollution synergistic effect capture capability analysis diagram according to the embodiment of the present application is shown in

[0069] Figure 3 A time-space decomposition and multi-factor correlation analysis diagram of transmission flux according to the embodiment of the present application is shown in

[0070] Figure 4 A real-time effect monitoring simulation diagram of the multi-pollution source linkage control strategy according to the embodiment of the present application is shown in DETAILED DESCRIPTION

[0071] In order to make the purpose, technical scheme and advantages of the embodiment of the present application clearer, the technical scheme in the embodiment of the present application will be described clearly and completely below in combination with the drawings in the embodiment of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0072] The technical scheme of the present application will be described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in some embodiments.

[0073] Figure 1 A flowchart of the method for collaborative governance and emission optimization and control of multiple pollution sources in the industrial park according to the embodiment of the present application is shown in Figure 1 as shown. The method comprises:

[0074] obtain real-time monitoring data and meteorological data of a plurality of pollution sources in an industrial park;

[0075] Based on the real-time monitoring data, the emission characteristic fingerprint data of each pollution source is generated by analyzing the characteristic pollutant component concentration and the ratio relationship of the characteristic pollutant component concentration of each pollution source, and the pollutant distribution weight coefficient is calculated according to the emission characteristic fingerprint data;

[0076] According to the emission characteristic fingerprint data, the meteorological data and the pollutant distribution weight coefficient, the gas-liquid phase state conversion rate, the chemical reaction rate and the settling rate of the pollutant are taken as calculation parameters by using a two-way fluid-solid coupling calculation method to generate pollutant transport path data, and the control partition of the industrial park is divided based on the pollutant transport path data;

[0077] According to the control partition and the pollutant transport path data, a pollutant migration and diffusion model is constructed, the pollutant superposition influence coefficient between each pollution source is calculated, a pollution source correlation strength matrix is generated, the optimal emission reduction proportion of each pollution source is calculated based on the pollution source correlation strength matrix and the pollutant distribution weight coefficient, the dynamic emission regulation and control parameters of each pollution source are calculated according to the optimal emission reduction proportion, the multi-pollution source linkage control strategy is established based on the dynamic emission regulation and control parameters and the pollutant transport path data, and the collaborative regulation and control instructions of each pollution source are generated.

[0078] In an alternative embodiment, based on the real-time monitoring data, the emission characteristic fingerprint data of each pollution source is generated by analyzing the characteristic pollutant component concentration and the ratio relationship of the characteristic pollutant component concentration of each pollution source, and the pollutant distribution weight coefficient is calculated according to the emission characteristic fingerprint data, which includes:

[0079] Obtain real-time monitoring data of a plurality of pollution sources in an industrial park, the real-time monitoring data including characteristic pollutant component concentration data, perform wavelet transform decomposition on the characteristic pollutant component concentration data to obtain first frequency band component concentration data and second frequency band component concentration data;

[0080] Construct a chemical reaction kinetics correction model based on the first frequency band component concentration data and the second frequency band component concentration data, calculate the pollutant direct reaction amount, the pollutant synergistic reaction amount and the pollutant decay amount by the chemical reaction kinetics correction model, and generate corrected component concentration data;

[0081] Perform time series analysis on the corrected component concentration data using an attention mechanism to extract the time distribution characteristics of the component concentration, generate time characteristic data, calculate the conversion relationship between the pollutant components according to the time characteristic data, and generate component conversion characteristic data;

[0082] Correlation between the pollutant components is calculated according to the time characteristic data and the component conversion characteristic data, a component action coefficient is generated, the time characteristic data and the component conversion characteristic data are weighted and fused by using the component action coefficient, and pollutant source emission characteristic fingerprint data is obtained;

[0083] Based on the pollutant source emission characteristic fingerprint data, the component concentration fluctuation law of each pollutant source in different time periods is analyzed, the pollutant concentration mean value, fluctuation amplitude and fluctuation period of each time period are calculated, and a pollutant distribution weight coefficient is obtained, which is used to represent the pollutant emission contribution of each pollutant source.

[0084] Exemplarily, real-time monitoring data of multiple pollutant sources in an industrial park is obtained. In this embodiment, a certain chemical industrial park is taken as an example, and 5 main pollution source points in the park are selected, including 2 chemical plants, 1 power plant, 1 garbage treatment plant and 1 sewage treatment plant. Through the online monitoring equipment installed at each pollution source, data is collected once an hour, continuously for 7 days, and characteristic pollutant component concentration data is obtained, including concentration values of pollutants such as SO2, nitrogen oxides (NOx), volatile organic compounds (VOCs), PM2.5 and NH3.

[0085] The obtained characteristic pollutant component concentration data is decomposed by wavelet transform. Using db4 wavelet basis function, the component concentration data of each pollution source is decomposed for 3 layers, and first frequency band component concentration data (low frequency part, representing the long-term change trend of pollutant concentration) and second frequency band component concentration data (high frequency part, representing the short-term fluctuation of pollutant concentration) are obtained. For example, for the SO2 concentration data of the chemical plant A, after wavelet decomposition, the first frequency band data reflects the concentration change caused by the production cycle, and the second frequency band data reflects the concentration fluctuation caused by sudden emission or meteorological condition change.

[0086] A chemical reaction kinetics correction model is constructed based on the decomposed frequency band component concentration data. The model considers the chemical reaction relationship between pollutants, including direct reaction, synergistic reaction and natural decay. For the direct reaction amount, taking the photochemical reaction of NOx and VOCs as an example, when the sunshine intensity is greater than 500 W / m 2 , and the temperature is higher than 25℃, the reaction rate of NOx and VOCs increases by about 30%. For the synergistic reaction amount, taking the formation of ammonium sulfate from SO2 and NH3 as an example, when the relative humidity is greater than 70%, the synergistic reaction amount of the two increases by about 25%. For the pollutant decay amount, taking VOCs as an example, according to the half-life period of different VOCs components (such as benzene series half-life period about 12 hours, alkane about 24 hours), the natural decay amount is calculated. Through these calculations, the corrected component concentration data is generated.

[0087] The revised component concentration data is analyzed using an attention mechanism for time series analysis. Specifically, a multi-head self-attention mechanism is used, with 8 attention heads and a dimension of 64 for each head to extract features of component concentrations at different time scales. For example, for the NOx emission data of a power plant, the attention mechanism can identify the emission pattern differences between weekdays and weekends, as well as the emission characteristics during morning and evening rush hours. In this way, time feature data is generated, including daily variation patterns, weekly variation patterns, and event characteristics, etc.

[0088] A pollutant component conversion network is constructed based on the time feature data. A graph neural network model is used, with each pollutant component as a node and the conversion relationship between components as an edge. The number of graph convolution layers is set to 3, with 128 hidden units in each layer and ReLU as the activation function. The conversion relationship between pollutant components is calculated through this network, such as the proportion of NOx converted to secondary organic aerosols under certain conditions, the rate of SO2 converted to sulfates, etc., thereby generating component conversion feature data.

[0089] The correlation between pollutant components is calculated based on the time feature data and the component conversion feature data. The Pearson correlation coefficient is used to calculate the correlation between each component, generating a component action coefficient matrix. For example, in the data of a waste treatment plant, the correlation coefficient between NH3 and H2S is 0.85, indicating that the two are highly correlated and may come from the same emission process.

[0090] The component action coefficient is used to weight and fuse the time feature data and the component conversion feature data. The time feature weight is set to 0.6 and the component conversion feature weight is set to 0.4, and the pollutant source emission feature fingerprint data is obtained by weighted summation. For example, the feature fingerprint of chemical plant B shows a VOCs / NOx ratio of 3.2, while the ratio for the power plant is 0.8, which can be used to distinguish different types of pollution sources.

[0091] Based on the pollutant source emission feature fingerprint data, the fluctuation rules of component concentrations at different time periods for each pollution source are analyzed. The mean, amplitude, and period of pollutant concentration are calculated for each time period to obtain the pollutant distribution weight coefficient. Taking pollution source A as an example, during the period of 9:00-17:00 on weekdays, the mean concentration of SO2 is 28 mg / m 3 , the amplitude is ±7 mg / m 3 , and the period is about 90 minutes; during the same period on non-working days, the mean concentration of SO2 is 12 mg / m3, the amplitude is ±3 mg / m3, and the period is about 120 minutes. Through these characteristic parameters, the SO2 distribution weight coefficients of pollution source A on weekdays and non-working days are calculated to be 0.65 and 0.25, respectively, indicating that the contribution of this pollution source to regional SO2 pollution on weekdays is significantly higher than on non-working days.

[0092] In the present embodiment, the feature pollutant concentration data is decomposed by wavelet transform to effectively extract pollution features in different frequency bands and improve the accuracy of data processing. In combination with a chemical reaction kinetics correction model, the direct reaction amount, synergistic reaction amount and decay amount of the pollutants can be corrected, so that the component concentration data is more consistent with the actual pollution diffusion situation. Time series analysis is performed using the attention mechanism to mine the time evolution characteristics of the pollutant concentration and enhance the ability to capture dynamic changes in pollution. The conversion relationship of the pollutant components is calculated using a graph neural network to construct a pollutant component conversion network and improve the analysis accuracy of the interaction between components. By calculating the component action coefficient and performing weighted fusion, the emission characteristic fingerprint data of the pollution source is efficiently extracted, thereby enhancing the traceability of the pollution source. Finally, by calculating the pollutant distribution weight coefficient, the calculation of the pollutant emission contribution is more scientific and reasonable, providing reliable decision-making basis for pollution tracing, environmental supervision and pollution control.

[0093] In an alternative embodiment, a chemical reaction kinetics correction model is constructed based on the first frequency band component concentration data and the second frequency band component concentration data, and the direct reaction amount of the pollutant, the synergistic reaction amount of the pollutant and the decay amount of the pollutant are calculated by the chemical reaction kinetics correction model to generate corrected component concentration data, including:

[0094] The first frequency band component concentration data and the second frequency band component concentration data are normalized to obtain standardized concentration data, and a reaction network structure is constructed based on the standardized concentration data to establish the conversion relationship between components and calculate the reaction intensity based on the concentration gradient between components.

[0095] A reaction rate equation of the chemical reaction kinetics correction model is established according to the reaction intensity, and the reaction rate equation includes a reactant concentration term, a reaction rate constant term and a correction coefficient term.

[0096] A frequency band coupling coefficient is obtained by solving the reaction rate equation using the standardized concentration data, a component concentration change matrix is established according to the frequency band coupling coefficient, the component concentration change matrix is substituted into the reaction rate equation to obtain a reaction rate value, and a direct reaction amount of the pollutant is calculated according to the product of the reaction rate value and the time step.

[0097] A component interaction matrix is constructed based on the component concentration change matrix, and the component interaction matrix is substituted into the reaction rate equation to calculate a synergistic reaction amount of the pollutant.

[0098] Environmental monitoring data is obtained, the correction coefficient term in the reaction rate equation is corrected according to the environmental monitoring data, and a decay amount of the pollutant is calculated.

[0099] The direct reaction amount of the pollutant and the synergistic reaction amount of the pollutant are added to the normalized concentration data, and the pollutant decay amount is subtracted, to generate the corrected component concentration data.

[0100] Exemplarily, the system acquires first frequency band component concentration data and second frequency band component concentration data. In actual application, the first frequency band can be the ultraviolet spectral region (such as 240-320 nm), and the second frequency band can be the visible spectral region (such as 380-780 nm). For example, a certain city air sample is monitored, and the concentration of SO2 in the ultraviolet frequency band is 120 μg / m 3 , and the concentration in the visible light frequency band is 105 μg / m 3 ; the concentration of NOx in the ultraviolet frequency band is 85 μg / m 3 , and the concentration in the visible light frequency band is 78 μg / m 3 ; the concentration of PM2.5 in the ultraviolet frequency band is 45 μg / m 3 , and the concentration in the visible light frequency band is 42 μg / m 3 .

[0101] The first frequency band component concentration data and the second frequency band component concentration data are normalized to obtain normalized concentration data. The normalization processing adopts the maximum-minimum normalization method, and maps each component concentration value to the 0-1 interval. Taking the above data as an example, the normalized concentration of SO2 in the ultraviolet frequency band is 1.0, and the normalized concentration in the visible light frequency band is 0.875; the normalized concentration of NOx in the ultraviolet frequency band is 0.708, and the normalized concentration in the visible light frequency band is 0.65; the normalized concentration of PM2.5 in the ultraviolet frequency band is 0.375, and the normalized concentration in the visible light frequency band is 0.35.

[0102] A reaction network structure is constructed based on the normalized concentration data, each component is taken as a node, and a material conversion relationship between components is established. For example, there is a photochemical reaction relationship between SO2 and NOx, an adsorption relationship between SO2 and PM2.5, and a catalytic relationship between NOx and PM2.5. The reaction intensity is calculated according to the concentration gradient between components, such as the ultraviolet frequency band reaction intensity between SO2 and NOx is 0.292 (i.e. 1.0 minus 0.708), and the visible light frequency band reaction intensity is 0.225 (i.e. 0.875 minus 0.65).

[0103] A reaction rate equation of a chemical reaction kinetics correction model is established according to the reaction intensity. The reaction rate equation includes a reactant concentration term, a reaction rate constant term, and a correction coefficient term. For the reaction between SO2 and NOx, the reactant concentration term in the reaction rate equation is the product of the normalized concentrations of the two, the initial value of the reaction rate constant term is set to 0.02, and the initial value of the correction coefficient term is set to 1.0.

[0104] The band coupling coefficient between SO2and NOxwas calculated to be 0.85 by iterative calculation, which indicated that the data in the ultraviolet band and the visible light band had high correlation. A component concentration change matrix was established according to the band coupling coefficient, which reflected the change rate of different component concentrations over time. For example, the concentration change rate of SO2was -0.018, and the concentration change rate of NOxwas -0.015, which indicated that the concentrations of both were reduced in the reaction process.

[0105] The reaction rate value was obtained by substituting the component concentration change matrix into the reaction rate equation. The calculated reaction rate value for the reaction of SO2and NOxwas 0.022. The direct reaction amount of pollutants was calculated according to the product of the reaction rate value and the time step. Assuming that the time step was 1 hour, the direct reaction amount of SO2was 0.022 x 1 = 0.022, which was converted to an actual concentration of about 2.64 μg / m 3 .

[0106] A component interaction matrix was constructed based on the component concentration change matrix. This matrix described the degree of mutual influence between components. For example, the influence coefficient of SO2on NOxwas 0.8, and the influence coefficient of NOxon SO2was 0.75. The synergistic reaction amount of pollutants was calculated by substituting the component interaction matrix into the reaction rate equation. Taking SO2as an example, its synergistic reaction amount was 0.015, which was converted to an actual concentration of about 1.8 μg / m3.

[0107] Environmental monitoring data such as temperature, humidity, and light intensity were obtained. In this embodiment, the environmental temperature was 25°C, the relative humidity was 65%, and the light intensity was 850 lux. The correction coefficient term in the reaction rate equation was corrected according to the environmental monitoring data. The correction coefficient increased by 0.2 for every 10°C increase in temperature, decreased by 0.1 for every 10% increase in relative humidity, and increased by 0.05 for every 100 lux increase in light intensity. Accordingly, the corrected coefficient was calculated to be 1.225. The reaction rate was recalculated using the corrected coefficient to obtain the pollutant decay amount. Taking SO2as an example, its decay amount was 0.008, which was converted to an actual concentration of about 0.96 μg / m 3 .

[0108] The pollutant direct reaction amount and the pollutant synergistic reaction amount were added to the normalized concentration data, and the pollutant decay amount was subtracted to generate the corrected component concentration data. Taking SO2as an example, the corrected normalized concentration was 1.0 + 0.022 + 0.015 - 0.008 = 1.029, which exceeded the normalized range and needed to be re-normalized. Finally, the corrected SO2concentration was obtained to be 123.48 μg / m 3 .

[0109] Through the above steps, a chemical reaction kinetics correction model was constructed based on multi-band component concentration data, and the corrected component concentration data was generated, thereby improving the accuracy of atmospheric pollutant concentration prediction.

[0110] Figure 2 This is a diagram illustrating the synergistic effect of pollutants in a specific embodiment of the present invention, showing the capture capability analysis. Figure 2 As shown in the figure, this graph compares the performance of different models in capturing the synergistic effects of pollutants. The X-axis represents different pollutant combinations, and the Y-axis represents the synergistic effect capture rate (%). Our proposed solution (◆) demonstrates excellent synergistic effect capture capability in all pollutant combinations, particularly in complex multi-component mixtures. In the SO2-NO2 combination, our solution achieves a capture rate of 85.7%, while the machine learning method (△) and the traditional box model (□) achieve only 62.3% and 48.5%, respectively. In the key photochemical reaction combination of NO2-O3-VOCs, our solution achieves a capture rate as high as 89.3%, 23.6 percentage points higher than the machine learning method. In the complex SO2-NO2-PM2.5-O3 four-component system, our solution maintains a high capture rate of 76.9%, while other methods significantly decrease to below 40%. These data fully demonstrate that the chemical reaction kinetics correction model constructed in our proposed solution can effectively identify and quantify complex interactions between pollutants, especially by constructing a component interaction matrix, successfully capturing synergistic reaction pathways that are difficult to express using traditional methods. In terms of the number of components, the capture capacity of each method generally decreases as the complexity of pollutant combinations increases, but the decrease in this technical solution is the smallest, maintaining a high level of capture rate.

[0111] The prior art generally uses simple statistical regression methods or empirical pollutant conversion models to estimate the reaction amount of pollutants, but these methods are difficult to accurately describe the complex reaction process of pollutants, ignore the dynamic coupling relationship between different pollutant components, and lead to deviations in the characterization of pollutant conversion rules, affecting the accuracy of pollution tracing and treatment. The present application normalizes the component concentration data and constructs a reaction network structure based on the standardized data, so that the material conversion relationship between different components can be represented in a structured manner, thereby improving the expression ability of pollutant reaction characteristics. Compared with existing methods, the present application introduces concentration gradient calculation of reaction strength, and characterizes the pollutant conversion process through a reaction rate equation, so that the calculation of the direct reaction amount of pollutants can combine the actual reaction kinetics characteristics, avoiding the simplification assumptions of traditional methods. Further, the present application calculates the synergistic reaction amount of pollutants through a component interaction matrix, which can more comprehensively consider the synergistic effect between pollutants compared to separately calculating the reaction amount of each pollutant, thereby improving the accuracy of reaction simulation. In addition, the present application uses environmental monitoring data to correct the reaction rate equation, so that the calculation of the pollutant decay amount can dynamically adapt to environmental changes, enhancing the adaptability and generalization ability of the model. Finally, the present application generates corrected component concentration data by comprehensively calculating the direct reaction amount, synergistic reaction amount and decay amount of pollutants, so that the simulation of the trend of pollutant concentration change is more consistent with the actual emission and conversion, thereby improving the scientificity and reliability of pollutant analysis.

[0112] In an alternative embodiment, according to the emission characteristic fingerprint data, meteorological data and pollutant distribution weight coefficient, the gas-liquid phase state conversion rate, chemical reaction rate and settling rate of the pollutants are calculated as calculation parameters by using a two-way fluid-structure coupling calculation method to generate pollutant transport path data, and the control partition of the industrial park is divided based on the pollutant transport path data, including:

[0113] The concentration distribution characteristics of different pollutant components in the emission characteristic fingerprint data are analyzed to calculate the mutual promotion strength and mutual inhibition strength between the pollutant components, and an interaction strength matrix between the components is constructed;

[0114] The interaction strength matrix between the components is combined with the pollutant distribution weight coefficient, the local pollutant concentration distribution characteristic value is calculated according to the temperature and humidity information in the meteorological data, the local concentration correction coefficient is obtained, the mass transfer resistance of the gas-liquid interface is corrected based on the local concentration correction coefficient, and the phase state conversion rate considering the interface resistance is calculated in combination with the interaction strength matrix between the components;

[0115] According to the phase transformation rate, the mass transfer flux of the gas-liquid interface is calculated, the mass transfer flux is combined with the local concentration correction coefficient, the chemical reaction rate considering the local distribution characteristics is calculated, and based on the chemical reaction rate and the phase transformation rate, the pollutant settling rate considering the interface mass transfer characteristics is calculated;

[0116] The phase transformation rate, chemical reaction rate and settling rate are substituted into the two-way fluid-structure coupling control equation, which contains the interaction term of gas phase pollutants and particulate matters, and an adaptive parameter adjustment mechanism is established according to the real-time change of environmental monitoring data, and the pollutant transport flux at each grid point is calculated by iterative solution;

[0117] The spatio-temporal variation characteristics of the transport flux are analyzed, the main transport channels are identified, and the pollution load index of each transport channel is calculated, and the pollutant transport path data is generated;

[0118] According to the pollutant transport path data, the diffusion risk coefficient of each grid point is calculated, and based on the pollution load index and the diffusion risk coefficient, the industrial park is divided into control zones.

[0119] Exemplarily, the emission characteristic fingerprint data is analyzed and processed. Taking a certain chemical industry park as an example, the emission characteristic fingerprint data of 10 kinds of main pollutants such as SO2, NOx and VOCs is collected, including concentration value, emission time, emission height and other information. By analyzing the concentration distribution characteristics of different pollutant components in these data, the mutual promotion intensity and mutual inhibition intensity between pollutant components are calculated. For example, SO2 and NOx will produce a synergistic effect under certain conditions, and the promotion intensity is 0.78; while VOCs and PM2.5 have a certain inhibitory effect, and the inhibition intensity is 0.32. Based on these numerical values, an interaction intensity matrix between components is constructed, which is a 10x10 square matrix, and the values on the diagonal are all 1, indicating the self-action intensity.

[0120] The pollution distribution weight coefficient is determined according to the importance of each pollutant to the environment, for example, the weight coefficient of SO2 is 0.85, the weight coefficient of NOx is 0.92, and the weight coefficient of VOCs is 0.76. At the same time, according to the temperature and humidity information in the meteorological data, the local pollutant concentration distribution characteristic value is calculated. In actual application, when the temperature is 25℃ and the relative humidity is 65%, the local concentration correction coefficient of SO2 is 1.12, the local concentration correction coefficient of NOx is 0.95, and the local concentration correction coefficient of VOCs is 1.23. Based on these local concentration correction coefficients, the mass transfer resistance of the gas-liquid interface is modified, for example, the modified gas-liquid interface mass transfer resistance coefficient is 0.85 times the original value. The phase transformation rate considering the interface resistance is calculated by combining the interaction intensity matrix between components.

[0121] Based on the calculated phase transformation rate, the mass transfer flux at the gas-liquid interface is calculated. The mass transfer flux is combined with the local concentration correction factor to calculate the chemical reaction rate considering the local distribution characteristics. Based on the chemical reaction rate and the phase transformation rate, the pollutant deposition rate is calculated considering the interface mass transfer characteristics, such as the deposition rate of SO2 is 0.008 m / s, NOx is 0.006 m / s, and VOCs is 0.004 m / s.

[0122] The phase transformation rate, chemical reaction rate, and deposition rate are substituted into the two-way fluid-structure coupling control equation. The equation includes the interaction term of gas phase pollutants and particulate matter. The industrial park is divided into grids with a size of 50 m x 50 m, totaling 80 x 60 grid points. An adaptive parameter adjustment mechanism is established based on the real-time changes of environmental monitoring data. When the deviation between monitoring data and calculation results exceeds 15%, the parameter is automatically adjusted, with an adjustment range of ±10% of the original parameter. The pollutant transport flux at each grid point is calculated by iterative solution, with an iteration number of 500 and a convergence accuracy of 10 -6 .

[0123] The spatiotemporal variation characteristics of the transport flux are analyzed, and the main transmission channels are identified and the pollution load index of each transmission channel is calculated. In practical application, 5 main transmission channels are identified, with a pollution load index of 0.87 for channel 1, 0.65 for channel 2, 0.59 for channel 3, 0.42 for channel 4, and 0.38 for channel 5. Based on these data, pollutant transport path data is generated, including transmission direction, transmission rate, and impact range, etc.

[0124] Based on the pollutant transport path data, the diffusion risk coefficient of each grid point is calculated. For example, the diffusion risk coefficient of the central area of the park is 0.92, the eastern area is 0.78, the western area is 0.65, the southern area is 0.53, and the northern area is 0.81. Based on the pollution load index and the diffusion risk coefficient, the industrial park is divided into control zones. Specifically, it is divided into four types of areas: strict control area (diffusion risk coefficient > 0.85 and pollution load index > 0.7), key control area (diffusion risk coefficient 0.7-0.85 or pollution load index 0.5-0.7), general control area (diffusion risk coefficient 0.5-0.7 and pollution load index 0.3-0.5), and buffer zone (diffusion risk coefficient < 0.5 and pollution load index < 0.3).

[0125] After the above method, the strict control area accounts for 15% of the area of the chemical industrial park, mainly distributed in the center and northeast of the park; the key control area accounts for 35%, mainly distributed around the strict control area; the general control area accounts for 40%, distributed more dispersedly; and the buffer area accounts for 10%, mainly located in the edge of the park. After implementing the zoning control measures for three months, the average concentration of SO2 in the park decreased by 23%, NOx decreased by 18%, and VOCs decreased by 25%, and the environmental air quality improved significantly, proving the effectiveness of the method.

[0126] In this embodiment, by constructing the component interaction strength matrix, the mutual promotion and inhibition relationship of different pollutant components can be quantified, and the description ability of complex interaction of pollutants can be improved. The local concentration correction coefficient is used to correct the mass transfer resistance of the gas-liquid interface, so that the calculation of phase transformation rate is more accurate, thereby optimizing the simulation of pollutant migration between gas and liquid phases. Combined with the calculation of the settling rate of pollutants, the settling process of pollutants is considered to have interface mass transfer characteristics, and the adaptability of the model to the actual environment is improved. Using the two-way fluid-structure coupling method, the interaction between gas-phase pollutants and particulate matter is introduced, the accurate simulation of pollutant diffusion dynamics is enhanced, and the response ability of the model to environmental changes is improved through the self-adaptive parameter adjustment mechanism. Finally, based on the pollutant transport path data, the main pollution transmission channels are accurately identified, and the pollution load index and diffusion risk coefficient are calculated, so as to realize the intelligent zoning control of the industrial park and improve the fine level of pollution control and the scientificity of decision-making.

[0127] In an alternative embodiment, the spatiotemporal variation characteristics of the transport flux are analyzed, the main transmission channels are identified, and the pollution load index of each transmission channel is calculated, and the generation of pollutant transport path data includes:

[0128] The pollutant concentration spatial gradient of the transport flux is calculated using the sliding time window method, the length of the sliding time window is dynamically adjusted according to the pollutant concentration change rate, the time sequence characteristics of the spatial gradient are analyzed, the gradient discrimination threshold is set according to the physical and chemical properties of the pollutant, the data sampling points are increased when the spatial gradient is greater than the gradient discrimination threshold, and the transport flux sampling data are obtained;

[0129] The transport flux sampling data are subjected to spatiotemporal decomposition to obtain the time evolution characteristics and spatial distribution characteristics of the transport flux, the spatiotemporal variation characteristics are subjected to multi-factor correlation analysis with the temperature field, humidity field and wind field, and a quantitative corresponding relationship between the spatiotemporal variation characteristics of the transport flux and the environmental field is established;

[0130] Based on the quantitative corresponding relationship, an environmental stress degree index is constructed, the environmental stress degree index represents the comprehensive influence of the environmental field on the diffusion, settling and accumulation of pollutants, and the pollution load index of each transmission channel is calculated by weighting and combining the environmental stress degree index and the transport flux.

[0131] constructing a strength-direction phase diagram of the transport flux, extracting the transport flux strength contour and direction field in the phase diagram, determining the main transport channel according to the continuity of the contour and direction field, and integrating the spatial position, transport flux and pollution load of the main transport channel to generate pollution transport path data.

[0132] The present application relates to a method for analyzing the spatiotemporal variation characteristics of transport flux, identifying the main transport channel and calculating the pollution load index. This method realizes efficient monitoring and evaluation of the pollution transport path through a series of steps, as follows:

[0133] The pollution concentration of the transport flux is analyzed using the sliding time window method. The core of this method is to dynamically adjust the length of the sliding time window according to the pollution concentration change rate. Specifically, a fixed length time window is initially set, and then the pollution concentration within the window is calculated at each time point. As the pollution concentration changes, if the concentration change rate exceeds the preset threshold, the length of the time window is increased accordingly to more comprehensively capture the characteristics of the concentration change. Through this dynamic adjustment, the spatial gradient of the pollution concentration can be effectively identified.

[0134] According to the physical and chemical properties of the pollutant, a gradient discrimination threshold is set. When the spatial gradient is greater than the threshold, the data sampling points are increased to obtain more accurate transport flux sampling data. This process ensures that in areas where the pollution concentration changes significantly, sufficient sample data can be obtained for subsequent analysis.

[0135] After obtaining the transport flux sampling data, temporal and spatial decomposition is performed to extract the temporal evolution characteristics and spatial distribution characteristics of the transport flux. At this time, the method of multi-factor correlation analysis is used to correlate the spatiotemporal variation characteristics of the transport flux with the temperature field, humidity field and wind field. By establishing a quantitative correspondence relationship, the influence of environmental factors on pollutant transport can be clearly determined. For example, by analyzing the changes in pollutant concentration in a specific area under different temperature, humidity and wind speed conditions, the comprehensive influence of environmental factors on pollutant diffusion, deposition and accumulation can be determined.

[0136] Based on the above quantitative correspondence relationship, an environmental stress index is constructed. This index takes into account the influence of environmental fields on pollutant diffusion, deposition and accumulation, and can effectively reflect the pressure of the environment on pollutant transport. By weighting and combining the environmental stress index with the transport flux, the pollution load index of each transport channel is calculated. The pollution load index can intuitively reflect the degree of influence of the channel on the environment, providing a scientific basis for subsequent environmental management.

[0137] After the calculation of the pollution load index, the intensity-direction phase diagram of the transport flux is constructed. This phase diagram can clearly show the main transport channel of the pollutants by extracting the contour and direction field of the transport flux intensity. According to the continuity of the contour and direction field, the spatial position of the main transport channel, the transport flux and the corresponding pollution load are determined. Finally, these information is integrated to generate the pollutant transport path data.

[0138] Taking the air pollutants in a city as the research object, the meteorological data and pollutant concentration data in the region are collected. In the preliminary analysis, the initial length of the sliding time window is set to 24 hours, and after dynamic adjustment, it is found that in some time periods, the pollutant concentration change rate is significant, and the time window length is increased to 48 hours. Through the spatio-temporal decomposition of these data, it is found that the pollutant concentration increases significantly under the conditions of high temperature, low humidity and strong wind.

[0139] In the multi-factor correlation analysis, it is found that for every 1 degree Celsius increase in temperature, the pollutant concentration increases by an average of 5 micrograms per cubic meter, and for every 1% decrease in humidity, the concentration increases by 3 micrograms per cubic meter. Through these data, the environmental stress index is calculated to be 0.75, indicating that the environmental pressure in this region is large. Combined with the transport flux, the pollution load index of the main transport channel is calculated to be 150, indicating that the channel has a significant impact on the environment.

[0140] Finally, through the construction of the intensity-direction phase diagram, the main pollutant transport channel is determined, and it is found that the pollutants mainly diffuse along the main roads and rivers of the city. Integrating these data, detailed pollutant transport path data is generated, providing a scientific basis for subsequent environmental management and pollution control.

[0141] In summary, through a series of systematic steps and methods, the present application can effectively analyze the spatio-temporal variation characteristics of the transport flux, identify the main transport channel, and calculate the pollution load index, providing important technical support for environmental monitoring and management.

[0142] Figure 3 The spatio-temporal decomposition and multi-factor correlation analysis of the transport flux of the embodiments of the present application are shown in the graph of FIG. Figure 3 The present technical solution constructs the environmental stress index through spatio-temporal decomposition and multi-factor correlation analysis, and considers the temperature field, humidity field and wind field comprehensively, and assigns weights of 0.35, 0.25 and 0.40 respectively. As shown in the graph, when the environmental stress is 0.2, the transport flux is 3.8 mg / m 2 / s (low stress area); when the environmental stress reaches 0.6, the transport flux rapidly rises to 9.6 mg / m 2 / s (critical point); at environmental stress degree 0.8, the transmission flux reaches the maximum value of 11.7 mg / m 2 / s; then under high stress environment (environmental stress degree 1.0), the transmission flux slightly decreases to 10.3 mg / m 2 / s. The multi-factor model of the technical solution achieves a determination coefficient (R2) of 0.93 and a mean square error (MSE) of 0.42. In comparison, the single-factor model (PMF source resolution method) only estimates a flux of 8.2 mg / m2 / s at the peak, which is 29.9% lower; the linear regression model (diffusion coefficient method) shows obvious linear deviation, and only predicts a flux of 6.4 mg / m 2 / s at the same stress degree, with a gap of 45.3% from the actual value. By considering the nonlinear interaction of environmental factors, the technical solution significantly improves the accuracy of transmission flux prediction.

[0143] The prior art usually relies on fixed time windows or simple interpolation methods to calculate the spatiotemporal distribution of pollutant transmission flux, but these methods are difficult to adapt to rapid changes in pollutant concentration, resulting in low accuracy in identifying transmission paths and affecting accurate characterization of the main pollution transmission channels. The present application calculates the spatial gradient of pollutant concentration by the sliding time window method, and dynamically adjusts the window length according to the concentration change rate, making the sampling strategy more flexible and able to accurately capture areas of rapid changes in pollutant concentration, improving the effectiveness of data sampling. Compared with existing methods, the present application introduces multi-factor correlation analysis of temperature field, humidity field and wind field when analyzing the spatiotemporal variation characteristics of transmission flux, establishing a quantitative correspondence between pollutant transmission flux and environmental factors, making the simulation of pollutant diffusion and deposition more consistent with actual environmental conditions, and avoiding the problem of insufficient consideration of external environmental factors in traditional methods. In addition, the present application constructs an environmental stress degree index to quantify the comprehensive influence of environmental fields on pollutant diffusion, deposition and accumulation, and calculates the pollution load index combined with the transmission flux, which can more comprehensively reflect the effect of environmental factors on pollutant migration compared to methods based solely on concentration distribution for pollution load assessment, improving the accuracy of pollution risk assessment. Finally, the present application uses the intensity-direction phase diagram of transmission flux, combined with the continuity of contour lines and direction field, to accurately identify the main pollution transmission channels, and integrates spatial position, flux and pollution load information to generate pollutant transmission path data, making the spatial diffusion process of pollutants clear and visualized, thereby improving the scientificity and accuracy of pollution monitoring and management.

[0144] In an alternative embodiment, a pollutant migration and diffusion model is constructed according to the control partition and pollutant transmission path data, a pollutant superposition influence coefficient between each pollution source is calculated, and a pollution source correlation strength matrix is generated, including:

[0145] According to the pollutant transmission path data, the control partition is meshed, the grid division density is adaptively adjusted based on the concentration gradient calculated by the pollutant concentration monitoring data, the mass conservation equation and the momentum conservation equation containing the chemical reaction rate of the pollutant, the gas-liquid phase conversion rate and the settling rate are established in the grid, and a pollutant migration and diffusion model is constructed;

[0146] The pollutant concentration field distribution data is obtained by solving the pollutant migration and diffusion model, the sensitivity coefficient representing the influence of the change of the pollutant source emission on the concentration of the receptor point is calculated based on the pollutant concentration field distribution data, the sensitivity coefficient is substituted into the nonlinear response function containing the gas phase oxidation, the liquid phase oxidation and the heterogeneous reaction, and the pollutant superposition influence coefficient is calculated;

[0147] The single-source transmission influence coefficient is calculated based on the pollutant superposition influence coefficient, the single-source transmission influence coefficient is input into the transfer closure algorithm for calculating the coupling transmission effect of the pollutant multi-source, the coupling transmission influence coefficient is obtained, and the single-source transmission influence coefficient and the coupling transmission influence coefficient are combined based on the time sequence sliding window to generate a pollution source correlation strength matrix.

[0148] Firstly, the control partition is meshed according to the pollutant transmission path data. Specifically, the geographical boundary data and the pollutant transmission path data of the control partition are obtained, and the control partition is preliminarily divided into uniform grids with a size of 500m x 500m. Then, the concentration gradient is calculated based on the pollutant concentration monitoring data, and the grid division density is adaptively adjusted. In the area with a large concentration gradient (such as a gradient value greater than 0.05 μg / m3 / km), the grid is refined to 100m x 100m; in the area with a medium concentration gradient (such as a gradient value between 0.01-0.05 μg / m3 / km), the grid is refined to 200m x 200m; and in the area with a small concentration gradient (such as a gradient value less than 0.01 μg / m 3 / km), the grid size of 500m x 500m is maintained.

[0149] In the divided grid, the mass conservation equation and the momentum conservation equation containing the chemical reaction rate of the pollutant, the gas-liquid phase conversion rate and the settling rate are established, and a pollutant migration and diffusion model is constructed. The mass conservation equation considers the emission, transmission, diffusion, chemical conversion and settling process of the pollutant. For PM2.5 pollutants, the chemical reaction rate is set to 0.01-0.05 / hour, the gas-liquid phase conversion rate is set to 0.005-0.02 / hour, the dry settling rate is set to 0.1-0.5 cm / s, and the wet settling rate is dynamically adjusted according to the precipitation intensity, ranging from 0.2-2.0 / hour. The momentum conservation equation considers the influence of the wind field, temperature field and pressure field on the transmission of the pollutant.

[0150] The pollutant concentration field distribution data is calculated by numerical solving method (e.g. finite difference method) using the constructed pollutant migration and diffusion model. The calculation time step is set to 1 hour, and the spatial resolution is consistent with the grid division. For example, in a certain typical case, the PM2.5 concentration in the simulation area ranges from 15 to 120 μg / m 3 , with the highest in the industrial area reaching 120 μg / m 3 , 60-90 μg / m 3 in the urban center area, and 15-40 μg / m 3 in the suburbs.

[0151] Based on the pollutant concentration field distribution data, the sensitivity coefficient representing the influence of the change in the emission amount of the pollution source on the concentration of the receptor point is calculated. The specific method is to simulate the increase of 10% and the decrease of 10% of the emission amount of each pollution source respectively, and calculate the change rate of the concentration of the receptor point. For example, when the emission amount of an industrial pollution source increases by 10%, the PM2.5 concentration of the receptor point 3 kilometers away increases by 4.2%, and the sensitivity coefficient of the pollution source to the receptor point is 0.42.

[0152] The sensitivity coefficient is substituted into the nonlinear response function including gas phase oxidation, liquid phase oxidation and heterogeneous reaction to calculate the pollutant superposition influence coefficient. The nonlinear response function takes into account the synergistic and antagonistic effects between different pollutants. For PM2.5, gas phase oxidation mainly considers the reaction with NOx, SO2 and VOCs, liquid phase oxidation mainly considers the reaction with soluble components in cloud droplets, and heterogeneous reaction mainly considers the reaction with the surface of particulate matter. Through these reaction paths, the pollutant superposition influence coefficient matrix is calculated. For example, in a certain case, two industrial pollution sources A and B 5 kilometers apart have a superposition influence coefficient of 1.2, indicating that their combined influence is 20% greater than the sum of their individual influences.

[0153] Based on the pollutant superposition influence coefficient, the single-source transmission influence coefficient is calculated. The single-source transmission influence coefficient represents the influence degree of a single pollution source on each receptor point. The calculation method is to multiply the emission intensity of the pollution source with the corresponding superposition influence coefficient, and then divide by the square of the distance between the receptor point and the pollution source. For example, the single-source transmission influence coefficient of pollution source A with an emission intensity of 100 tons / year to a receptor point 2 kilometers away is 15 μg / m 3 .

[0154] The single-source transmission influence coefficient is input into a transfer closure algorithm for calculating the coupling transmission effect of multiple sources to obtain a coupling transmission influence coefficient. The transfer closure algorithm considers multiple interactions of the pollutants in the transmission process through iterative calculation. For example, in a certain area with five main pollution sources, the coupling transmission influence coefficient matrix calculated by the transfer closure algorithm shows that the coupling transmission influence coefficient between pollution sources A and C is 2.3, which is much higher than the direct influence coefficient of 1.5, indicating that there is a significant indirect coupling effect.

[0155] The single-source transmission influence coefficient and the coupling transmission influence coefficient are combined based on a time series sliding window to generate a pollution source correlation strength matrix. The size of the sliding window is set to 24 hours, and the step is 6 hours. The influence coefficients in each time window are weighted and averaged, and the weight is proportional to the pollutant concentration. For example, during the winter heating period, the correlation strength matrix of the five main pollution sources in an industrial area shows that the correlation strength between pollution sources A and B is 0.85 (full score is 1), indicating that there is a strong mutual influence between them; while the correlation strength between pollution sources A and E is only 0.12, indicating that the mutual influence between them is weak.

[0156] Through the above steps, the whole process of constructing a pollutant migration and diffusion model based on control partitioning and pollutant transmission path data, calculating the pollutant superposition influence coefficient between each pollution source, and generating a pollution source correlation strength matrix is completed, providing a scientific basis for regional air pollution joint prevention and control.

[0157] In this embodiment, through adaptive grid division based on pollutant concentration gradient, the calculation accuracy of the area with rapid change of pollutant concentration is improved, and the problems of insufficient resolution or waste of computing resources caused by the existing fixed grid method are avoided. By using the mass conservation and momentum conservation equations containing chemical reaction rate, phase transformation rate and settling rate, a pollutant migration and diffusion model more consistent with the actual physical process is established, so that the transmission dynamics of pollutants under different environmental conditions can be accurately simulated. Compared with the traditional method which ignores the nonlinearity of pollutant reaction, the accuracy and adaptability of the model are enhanced. In addition, the application calculates the pollutant superposition influence coefficient based on the sensitivity coefficient of the pollutant concentration field and combined with the nonlinear response function, which can more comprehensively reflect the change of the pollutant concentration under the joint action of multiple pollution sources compared with the single pollution source analysis method. Further, the coupling transmission influence coefficient is calculated by the transfer closure algorithm, and the data of different time periods are combined by using the sliding window, so that the pollution source correlation strength matrix can dynamically reflect the transmission relationship and time evolution characteristics of the pollution sources, thereby improving the analysis ability of the pollution causes and providing a more scientific decision basis for accurate source tracing and regional pollution control.

[0158] In an alternative embodiment, the optimal emission reduction proportion of each pollution source is calculated based on the pollution source correlation intensity matrix and the pollutant distribution weight coefficient, the dynamic emission regulation parameter of each pollution source is calculated according to the optimal emission reduction proportion, the multi-pollution source linkage control strategy is established based on the dynamic emission regulation parameter and the pollutant transmission path data, and the collaborative regulation instruction of each pollution source is generated, including:

[0159] The pollution source correlation intensity matrix and the pollutant distribution weight coefficient are input into a nonlinear programming model, the nonlinear programming model includes an objective function of maximizing pollutant reduction and minimizing economic cost, and the optimal emission reduction proportion of each pollution source is obtained by solving the nonlinear programming model through a genetic algorithm;

[0160] A state space model including pollution source characteristics, meteorological conditions and environmental capacity is established according to the optimal emission reduction proportion, a Kalman filtering algorithm is used to estimate the system state in real time, and the state estimation result is input into a model predictive controller to calculate the dynamic emission regulation parameter of each pollution source;

[0161] A pollution source weighted directed graph is constructed based on the dynamic emission regulation parameter and the pollutant transmission path data, a community discovery algorithm is used to identify a strongly correlated pollution source group in the weighted directed graph, a differentiated group control strategy is generated according to the dynamic emission regulation parameter of each group, and a multi-pollution source linkage control strategy is established;

[0162] A hierarchical control structure is constructed based on the multi-pollution source linkage control strategy, a feedback compensation mechanism is set in the hierarchical control structure, actual pollutant concentration change data is input into the feedback compensation mechanism for dynamic correction, and a collaborative regulation instruction of each pollution source is generated.

[0163] For example, the optimal emission reduction proportion of each pollution source is calculated based on the pollution source correlation intensity matrix and the pollutant distribution weight coefficient. The pollution source correlation intensity matrix represents the degree of mutual influence between different pollution sources, which can be constructed by historical monitoring data and diffusion model. For example, there are 5 main pollution sources in a certain area, and the correlation intensity matrix obtained by analyzing historical data shows that the correlation intensity between pollution source 1 and pollution source 2 is 0.75, indicating that the emission change of pollution source 1 will have a strong impact on the area where pollution source 2 is located. The pollutant distribution weight coefficient reflects the contribution of each pollutant to environmental quality, for example, the weight coefficient of PM2.5 is 0.4, the weight coefficient of SO2 is 0.3, and the weight coefficient of NOx is 0.3.

[0164] The above data is input into a nonlinear programming model, which includes two objective functions: maximum pollutant reduction and minimum economic cost. The objective of maximum pollutant reduction is to maximize the reduction of pollutant concentration in the region; the objective of minimum economic cost is to minimize the economic cost of emission reduction measures. The model constraints include: the upper limit of the emission reduction proportion of each pollution source is not more than 80%, the improvement of regional environmental quality is not less than 20%, and the economic impact is not more than 2% of the regional GDP, etc.

[0165] The genetic algorithm is used to solve the nonlinear programming model, and the algorithm parameters are set as follows: population size 100, maximum iteration number 500, crossover probability 0.8, and mutation probability 0.1. Through iterative optimization, the optimal emission reduction proportion of each pollution source is obtained, such as the optimal emission reduction proportion of pollution source 1 is 35%, pollution source 2 is 42%, pollution source 3 is 28%, pollution source 4 is 50%, and pollution source 5 is 20%.

[0166] A state space model is established, which includes pollution source characteristics, meteorological conditions and environmental capacity. Pollution source characteristics include emission height, emission temperature, emission rate and other parameters; meteorological conditions include wind direction, wind speed, atmospheric stability, etc.; environmental capacity is determined based on regional environmental quality standards.

[0167] Taking an industrial park as an example, pollution source 1 is a thermal power plant, with an emission height of 120 meters, an emission temperature of 85°C, and an initial emission rate of SO2 200kg / h, NOx 150kg / h; the meteorological conditions of the day are northeast wind, wind speed 3.5m / s, and atmospheric stability D level; the regional SO2 environmental capacity is 150μg / m 3 , and the NOx environmental capacity is 100μg / m 3 .

[0168] Kalman filtering algorithm is used to estimate the system state in real time, process noise and observation noise, and improve the accuracy of state estimation. The initial parameters of the algorithm are set as follows: the initial value of the state estimation error covariance matrix P is the unit matrix, the process noise covariance matrix Q is set to 0.01, and the observation noise covariance matrix R is set to 0.05. Through continuous observation data update, the optimal estimation value of the system state is obtained.

[0169] The state estimation results are input into the model predictive controller, the prediction time domain is set to 24 hours, the control time domain is set to 8 hours, and the sampling period is 1 hour. The controller calculates the dynamic emission control parameters of each pollution source according to the predicted pollutant diffusion trend and environmental capacity constraints. For example, the SO2 emission control parameters of pollution source 1 in the next 8 hours are: 130kg / h, 120kg / h, 140kg / h, 150kg / h, 130kg / h, 120kg / h, 110kg / h, 100kg / h.

[0170] A multi-pollution source linkage control strategy is established based on dynamic emission regulation parameters and pollutant transmission path data. A weighted directed graph of pollution sources is constructed, in which nodes represent pollution sources and the weight of edges represents the intensity of pollutant transmission, and the direction represents the transmission path. For example, the weight of the edge from pollution source 1 to pollution source 3 is 0.65, indicating that 65% of the pollutants emitted by pollution source 1 will affect the area where pollution source 3 is located.

[0171] Optionally, the Louvain community discovery algorithm is used to identify the strongly correlated pollution source groups in the weighted directed graph, and the algorithm resolution parameter is set to 1.0. Through algorithm analysis, the five pollution sources are divided into two groups: group 1 contains pollution sources 1, 2, and 3, and group 2 contains pollution sources 4 and 5. The average internal correlation strength of group 1 is 0.72, the average internal correlation strength of group 2 is 0.68, and the average correlation strength between groups is 0.31.

[0172] According to the dynamic emission regulation parameters of each group, a differentiated group control strategy is generated. For group 1, which contains a high-emission thermal power plant, a "gradient emission reduction" strategy is adopted, i.e., reducing emissions in advance before the pollution peak period; for group 2, which is mainly intermittent emission sources, a "peak-shifting emission" strategy is adopted to avoid high-intensity emissions at the same time.

[0173] Based on the multi-pollution source linkage control strategy, a hierarchical control structure is constructed, including three levels of regional layer, group layer, and source point layer. The regional layer is responsible for overall target decomposition, the group layer coordinates the linkage of pollution sources within the group, and the source point layer executes specific control instructions. A feedback compensation mechanism is set in the hierarchical control structure to collect pollutant concentration data from environmental monitoring stations in real time, and when the actual concentration deviates from the expected target by more than 10%, compensation adjustment is triggered.

[0174] Taking a certain day as an example, monitoring found that the regional PM2.5 concentration was 75 μg / m 3 , which was 15% higher than the expected target of 65 μg / m 3 . The feedback compensation mechanism automatically adjusts the emission parameters of each pollution source: the SO2 emission of pollution source 1 is adjusted from the original plan of 130 kg / h to 110 kg / h, a decrease of 15.4%; the NOx emission of pollution source 2 is adjusted from the original plan of 90 kg / h to 75 kg / h, a decrease of 16.7%. Through this dynamic adjustment, the final coordinated control instructions for each pollution source are generated, achieving effective control of regional pollutant concentration.

[0175] Through the above technical solutions, the present application realizes precise coordinated control of multiple pollution sources, maximizes the improvement of environmental quality while ensuring economic development, and has significant practical value.

[0176] In this embodiment, a nonlinear programming model is constructed by a pollution source correlation strength matrix and a pollutant distribution weight coefficient to optimize the emission reduction ratio, maximize the pollutant reduction effect, and reduce the emission reduction cost. Compared with the traditional fixed proportion emission reduction method, the method is more targeted and adaptive. The Kalman filtering algorithm is combined with the state space model to realize real-time estimation of the emission state of the pollution source, and the model predictive controller is used to dynamically adjust the emission control parameters, so that the pollution control strategy can adapt to the real-time changes of the environmental conditions, and the problem of slow response to sudden pollution events in the existing method is overcome. In addition, the weighted directed graph is constructed by using the pollutant transmission path data, and the strongly correlated pollution source group is identified by using the community discovery algorithm, so that the pollution control can be differentiated based on the actual influence relationship of the pollution sources, and the limitation of independent control of each pollution source in the traditional method is avoided. Finally, the pollution control strategy is dynamically corrected by using the hierarchical control structure and the feedback compensation mechanism, so that the control instruction can adapt to the actual changes of the pollutant concentration, thereby improving the control precision and ensuring the stability and continuity of the pollution control effect, and optimizing the regional pollution prevention and control decision.

[0177] In an optional embodiment, a pollution source weighted directed graph is constructed based on the dynamic emission control parameters and the pollutant transmission path data, a community discovery algorithm is used to identify a strongly correlated pollution source group in the weighted directed graph, a differentiated group control strategy is generated according to the dynamic emission control parameters of each group, and a multi-pollution source linkage control strategy is established, including:

[0178] The dynamic emission control parameters are taken as node characteristic values, the node connection relationship is constructed based on the pollutant transmission path data, the transmission amount between nodes is updated in an exponential decay manner to obtain edge characteristic values, and a pollution source weighted directed graph is constructed according to the node characteristic values and the edge characteristic values;

[0179] The actual transmission strength and the network reference transmission strength between nodes are calculated based on the pollution source weighted directed graph, the actual transmission strength and the network reference transmission strength are input into a community discovery algorithm, and an initial group division scheme is obtained by iteratively optimizing a group division index;

[0180] The transmission distance coefficient and the meteorological similarity coefficient between groups are calculated based on the initial group division scheme, the transmission distance coefficient and the meteorological similarity coefficient are combined with the group division index to construct a group optimization criterion, the initial group division scheme is optimized according to the group optimization criterion to obtain a target pollution source group, a group control model including a pollutant reduction amount and a control cost is constructed for the target pollution source group, and a pollution source group control target is obtained by solving the group control model;

[0181] The connection degree coefficient, the transmission flux coefficient and the position correlation coefficient of the nodes in each pollution source group are calculated, the node priority coefficient is obtained by combining the connection degree coefficient, the transmission flux coefficient and the position correlation coefficient, and the nodes in the pollution source group are ranked based on the node priority coefficient to obtain an intra-group ranking scheme;

[0182] A group game model containing pollution reduction amount and regulation cost is constructed according to the intra-group ranking scheme and the pollution source group regulation target, a differentiated group control strategy is obtained by solving the group game model, and a regulation correction amount is calculated by collecting real-time monitoring data, the regulation correction amount is input into a compensation model to adjust the differentiated group control strategy in real time, and a multi-pollution source linkage control strategy is established.

[0183] Illustratively, a pollution source weighted directed graph is constructed based on dynamic emission regulation parameters and pollutant transmission path data. Specifically, each pollution source is taken as a node in the graph, and the dynamic emission regulation parameters are taken as the node characteristic values. These parameters include the emission intensity, emission height, flue gas temperature, etc. of the pollution source. For example, the emission intensity of a certain steel plant is 200 tons / year, the emission height is 80 meters, and the flue gas temperature is 150°C, which together constitute the characteristic values of the node of this pollution source.

[0184] Based on the pollutant transmission path data, the connection relationship between nodes is constructed, i.e. the edges in the directed graph are determined. For example, according to the atmospheric diffusion model, the pollutant contribution rate of A pollution source to the area where B pollution source is located is calculated to be 15%, and a directed edge from A to B is established. The transmission amount between nodes is updated in an exponential decay manner, i.e. as time passes or distance increases, the transmission influence decreases according to an exponential law. For example, when the transmission distance is 10 kilometers, the transmission coefficient may be 0.8; when the distance increases to 20 kilometers, the transmission coefficient may decrease to 0.64. These transmission amounts are taken as the characteristic values of the edges, and finally a complete pollution source weighted directed graph is formed.

[0185] Based on the constructed pollution source weighted directed graph, strongly correlated pollution source groups are identified. The actual transmission intensity between nodes is calculated, such as the actual transmission intensity of A pollution source to B pollution source, which is the emission amount of A multiplied by the transmission coefficient, to obtain a specific value such as 30 tons / year. At the same time, the network reference transmission intensity is calculated, i.e. the average transmission intensity between all nodes in the entire network, such as 20 tons / year, which is the total transmission amount of 2000 tons / year in a network with 100 nodes.

[0186] The actual transmission intensity and the network reference transmission intensity are input into a community discovery algorithm, such as the Louvain algorithm or the InfoMap algorithm, and an initial group division scheme is obtained by iteratively optimizing the group division index (such as the modularity Q value). For example, the initial division may divide 100 pollution sources into 5 groups, with tight connections within each group and relatively sparse connections between groups.

[0187] Based on the initial group division scheme, the inter-group transmission distance coefficient and meteorological similarity coefficient are calculated. The transmission distance coefficient reflects the geographical distance relationship between groups, such as the average distance between group A and group B is 15 kilometers, and the corresponding transmission distance coefficient is 0.7. The meteorological similarity coefficient reflects the similarity of the meteorological conditions in the area where the group is located, such as the wind direction similarity between group C and group D is 85%, and the wind speed similarity is 90%, and the comprehensive meteorological similarity coefficient is 0.87.

[0188] The transmission distance coefficient and meteorological similarity coefficient are combined with the group division index to construct the group optimization criterion. For example, the optimization criterion can be set as: modularity Q value x 0.5 + transmission distance coefficient x 0.3 + meteorological similarity coefficient x 0.2. According to this criterion, the initial group division scheme is optimized, and the original 5 groups may be adjusted to 6 groups to obtain the target pollution source group.

[0189] For the target pollution source group, a group control model is constructed, which includes the amount of pollutant reduction and the control cost. For example, group E needs to reduce the amount of pollutant emission by 100 tons, and the total control cost is not more than 5 million yuan. By solving the model, the pollution source group control target is obtained, such as group E needs to reduce sulfur dioxide by 80 tons and nitrogen oxides by 120 tons, with a total cost of 4.5 million yuan.

[0190] The connection degree coefficient, transmission flux coefficient and location correlation coefficient of each node in the pollution source group are calculated. The connection degree coefficient reflects the connection tightness of the node in the group, such as the connection number of a node with other nodes in the group is 8, and the average connection number in the group is 5, then the connection degree coefficient of the node is 1.6. The transmission flux coefficient reflects the total amount of pollutant transmission of the node, such as the outward transmission amount of a node is 40 tons / year, the receiving transmission amount is 20 tons / year, the total flux is 60 tons / year, and the average flux of the group is 40 tons / year, then the transmission flux coefficient of the node is 1.5. The location correlation coefficient reflects the importance of the node in geographical position, such as a node is located in the upwind area, which has significant influence on the downwind, and its location correlation coefficient may be 1.8.

[0191] The connection degree coefficient, transmission flux coefficient and location correlation coefficient are combined to obtain the node priority coefficient. For example, the priority coefficient can be set as: connection degree coefficient x 0.4 + transmission flux coefficient x 0.4 + location correlation coefficient x 0.2. Based on the node priority coefficient, the nodes in the pollution source group are classified, such as the priority coefficient greater than 1.5 is a first-level node, 1.0-1.5 is a second-level node, and less than 1.0 is a third-level node, thereby obtaining the classification scheme in the group.

[0192] According to the hierarchical scheme within the group and the pollution source group control target, a group game model containing pollutant reduction amount and control cost is constructed. In this model, different levels of nodes bear different proportion of emission reduction tasks, such as first-level nodes bear 50% of emission reduction amount, second-level nodes bear 30%, and third-level nodes bear 20%. The game model is solved to obtain differentiated group control strategies, such as first-level nodes average 40 tons of emission reduction, second-level nodes average 20 tons of emission reduction, and third-level nodes average 10 tons of emission reduction.

[0193] The real-time monitoring data is collected to calculate the control correction amount, for example, the real-time monitoring finds that the pollutant concentration in a certain area exceeds 20%, and needs to increase the emission reduction amount by 15 tons. The control correction amount is input into the compensation model to adjust the differentiated group control strategy in real time, such as adjusting the emission reduction amount of first-level nodes from 40 tons to 46 tons, adjusting the emission reduction amount of second-level nodes from 20 tons to 23 tons, and adjusting the emission reduction amount of third-level nodes from 10 tons to 11.5 tons, and finally establishing a complete multi-pollution source linkage control strategy.

[0194] Through the above method, the group division based on the correlation of pollution sources, the hierarchical control based on the importance of nodes, and the dynamic adjustment based on real-time data are realized, which provides scientific and effective technical support for regional air pollution joint prevention and control.

[0195] Figure 4 The figure shows the real-time air quality change trend, and the technical scheme (solid line) keeps the pollutant concentration below 100 μg / m 3 all day, and the highest reaches 95 μg / m 3 ; while the traditional method (dashed line) exceeds the standard during 8-20 hours, and the highest reaches 135 μg / m 3 . The technical scheme reduces the pollutant concentration by an average of 28.3% in the key period. Through the real-time adjustment of the differentiated group control strategy by the compensation model, the technical scheme realizes dynamic linkage control based on real-time monitoring data, and significantly improves the pollutant control efficiency and the environmental target achievement rate.

[0196] The prior art usually adopts single-source independent regulation or fixed proportion reduction in multi-pollution source control, ignores the mutual influence between pollution sources, and leads to lack of precision of the control strategy and limitation of the emission reduction effect. The application accurately describes the transmission relationship between pollution sources by constructing a weighted directed graph of pollution sources, combining pollution transmission path data and dynamic emission regulation parameters, and can more finely depict the pollution diffusion characteristics and improve the accuracy of pollution source identification compared with traditional methods. The community discovery algorithm is used to optimize the division of pollution source groups, and the group optimization criteria are adjusted based on the transmission distance and meteorological similarity, so that the pollution source classification is more scientific and the errors caused by single index division in traditional methods are overcome. Further, the group game model is used to develop differentiated group control strategies based on the pollutant reduction amount and the regulation cost, which avoids the one-size-fits-all emission reduction method and improves the flexibility of pollution control. At the same time, the application calculates the regulation correction amount by using real-time monitoring data, and dynamically adjusts the control strategy through the compensation model, so that the regulation scheme can adapt to environmental changes and improve the real-time response capability of pollution control. Compared with the prior art, the application realizes the upgrade from static reduction to dynamic optimization, improves the precision and adaptability of multi-pollution source collaborative regulation, reduces the governance cost, and ensures the stability and integrity of the pollutant reduction effect.

[0197] A second aspect of the embodiments of the application,

[0198] An electronic device is provided, comprising:

[0199] A processor;

[0200] A memory for storing processor-executable instructions;

[0201] The processor is configured to invoke the instructions stored in the memory to perform the method described above.

[0202] A third aspect of the embodiments of the application,

[0203] A computer-readable storage medium is provided, which stores computer program instructions, and the computer program instructions are executed by a processor to implement the method described above.

[0204] The application can be a method, device, system and / or computer program product. The computer program product can include a computer readable storage medium on which is loaded computer readable program instructions for executing various aspects of the application.

[0205] 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 the present application; although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions recorded in the above embodiments can be modified, or some or all of the technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for multi-pollution source collaborative governance and emission optimization regulation and control in an industrial park, characterized in that, The method comprises the following steps: acquiring real-time monitoring data and meteorological data of a plurality of pollution sources in an industrial park; based on the real-time monitoring data, by analyzing the concentration of characteristic pollutant components and the ratio of the concentration of characteristic pollutant components of each pollution source, generating the emission characteristic fingerprint data of each pollution source, and calculating the pollutant distribution weight coefficient according to the emission characteristic fingerprint data; analyzing the concentration distribution characteristics of different pollutant components in the emission characteristic fingerprint data, and calculating the mutual promotion intensity and mutual inhibition intensity between the pollutant components to construct the component interaction strength matrix; combining the component interaction strength matrix with the pollutant distribution weight coefficient, calculating the local pollutant concentration distribution eigenvalue according to the temperature and humidity information in the meteorological data, obtaining the local concentration correction coefficient, modifying the mass transfer resistance of the gas-liquid interface based on the local concentration correction coefficient, and calculating the phase transformation rate considering the interface resistance by combining the component interaction strength matrix; calculating the mass transfer flux of the gas-liquid interface according to the phase transformation rate, combining the mass transfer flux with the local concentration correction coefficient to calculate the chemical reaction rate considering the local distribution characteristics, and calculating the pollutant settling rate considering the interface mass transfer characteristics based on the chemical reaction rate and the phase transformation rate; substituting the phase transformation rate, chemical reaction rate and settling rate into the two-way fluid-solid coupling control equation, which contains the interaction term of gas phase pollutants and particulate matter, and establishing an adaptive parameter adjustment mechanism according to the real-time change of environmental monitoring data, and calculating the pollutant transport flux at each grid point by iterative solution; analyzing the spatiotemporal variation characteristics of the transport flux, identifying the main transmission channels and calculating the pollution load index of each transmission channel to generate the pollutant transport path data; calculating the diffusion risk coefficient of each grid point according to the pollutant transport path data, and dividing the industrial park into control zones based on the pollution load index and the diffusion risk coefficient; constructing a pollutant migration and diffusion model according to the control zones and the pollutant transport path data, calculating the pollutant superposition influence coefficient between each pollution source, generating the pollution source correlation strength matrix, calculating the optimal emission reduction proportion of each pollution source based on the pollution source correlation strength matrix and the pollutant distribution weight coefficient, calculating the dynamic emission control parameters of each pollution source according to the optimal emission reduction proportion, establishing a multi-pollution source linkage control strategy based on the dynamic emission control parameters and the pollutant transport path data, and generating the collaborative control instructions of each pollution source.

2. The method of claim 1, wherein, Based on the real-time monitoring data, by analyzing the concentration of characteristic pollutant components and the ratio of the concentration of characteristic pollutant components of each pollution source, generating the emission characteristic fingerprint data of each pollution source, and calculating the pollutant distribution weight coefficient comprises: acquiring real-time monitoring data of a plurality of pollution sources in an industrial park, the real-time monitoring data including characteristic pollutant component concentration data, and performing wavelet transform decomposition on the characteristic pollutant component concentration data to obtain first frequency band component concentration data and second frequency band component concentration data; constructing a chemical reaction kinetics correction model based on the first frequency band component concentration data and the second frequency band component concentration data, calculating a pollutant direct reaction amount, a pollutant synergistic reaction amount and a pollutant attenuation amount through the chemical reaction kinetics correction model, and generating corrected component concentration data; performing time series analysis on the corrected component concentration data using an attention mechanism to extract time distribution characteristics of the component concentration, generating time characteristic data, calculating a conversion relationship between pollutant components according to the time characteristic data, and generating component conversion characteristic data; calculating a correlation between pollutant components according to the time characteristic data and the component conversion characteristic data, generating a component action coefficient, and performing weighted fusion on the time characteristic data and the component conversion characteristic data using the component action coefficient to obtain pollutant source emission characteristic fingerprint data; based on the pollutant source emission characteristic fingerprint data, analyzing the component concentration fluctuation law of each pollutant source in different time periods, calculating the pollutant concentration mean value, fluctuation amplitude and fluctuation period of each time period, and obtaining a pollutant distribution weight coefficient, which is used to represent the pollutant emission contribution of each pollutant source.

3. The method of claim 2, wherein, constructing a chemical reaction kinetics correction model based on the first frequency band component concentration data and the second frequency band component concentration data, calculating a pollutant direct reaction amount, a pollutant synergistic reaction amount and a pollutant attenuation amount through the chemical reaction kinetics correction model, and generating corrected component concentration data includes: normalizing the first frequency band component concentration data and the second frequency band component concentration data to obtain standardized concentration data, constructing a reaction network structure based on the standardized concentration data, establishing a material conversion relationship between components as nodes, and calculating a concentration gradient between components to obtain a reaction intensity; establishing a reaction rate equation of the chemical reaction kinetics correction model according to the reaction intensity, the reaction rate equation including a reactant concentration term, a reaction rate constant term and a correction coefficient term; solving the reaction rate equation using the standardized concentration data to obtain a frequency band coupling coefficient, establishing a component concentration change matrix according to the frequency band coupling coefficient, substituting the component concentration change matrix into the reaction rate equation to obtain a reaction rate value, and calculating a pollutant direct reaction amount according to the product of the reaction rate value and a time step; constructing a component interaction matrix based on the component concentration change matrix, substituting the component interaction matrix into the reaction rate equation to calculate a pollutant synergistic reaction amount; obtaining environmental monitoring data, correcting the correction coefficient term in the reaction rate equation according to the environmental monitoring data, and calculating a pollutant attenuation amount; adding the pollutant direct reaction amount and the pollutant synergistic reaction amount to the standardized concentration data and subtracting the pollutant attenuation amount to generate corrected component concentration data.

4. The method of claim 1, wherein, analyzing the spatiotemporal variation characteristics of the transmission flux, identifying the main transmission channels and calculating the pollution load index of each transmission channel, and generating pollutant transmission path data includes: The spatial gradient of the pollutant concentration of the transmission flux is calculated by using a sliding time window method, a length of the sliding time window is dynamically adjusted according to a variation rate of the pollutant concentration, a time sequence characteristic of the spatial gradient is analyzed, a gradient discrimination threshold is set according to physical and chemical characteristics of the pollutant, a data sampling point is increased when the spatial gradient is greater than the gradient discrimination threshold, and transmission flux sampling data are obtained; The transmission flux sampling data are subjected to space-time decomposition, time evolution characteristics and spatial distribution characteristics of the transmission flux are obtained, the time and space variation characteristics are subjected to multi-factor correlation analysis with a temperature field, a humidity field and a wind field, and a quantitative corresponding relationship between the time and space variation characteristics of the transmission flux and the environmental field is established; An environmental stress degree index is constructed based on the quantitative corresponding relationship, the environmental stress degree index represents a comprehensive influence of the environmental field on diffusion, deposition and accumulation of the pollutant, the environmental stress degree index and the transmission flux are subjected to weighted combination, and a pollution load index of each transmission channel is calculated; An intensity-direction phase diagram of the transmission flux is constructed, transmission flux intensity isosurfaces and a direction field in the phase diagram are extracted, a main transmission channel is determined according to continuity of the isosurfaces and the direction field, and pollution source correlation strength matrixes are generated by integrating spatial positions, transmission fluxes and pollution loads of the main transmission channel.

5. The method of claim 1, wherein, A pollutant migration and diffusion model is constructed according to the pollution control partition and the pollutant transmission path data, a pollutant superposition influence coefficient between pollution sources is calculated, and the pollution source correlation strength matrixes are generated, including: The pollution control partition is subjected to grid division according to the pollutant transmission path data, a grid division density is adaptively adjusted by using a concentration gradient calculated based on pollutant concentration monitoring data, mass conservation equations and momentum conservation equations including a pollutant chemical reaction rate, a gas-liquid phase conversion rate and a deposition rate are established in the grid, and the pollutant migration and diffusion model is constructed; Pollutant concentration field distribution data are obtained by solving the pollutant migration and diffusion model, a sensitivity coefficient representing an influence of a change in an emission amount of a pollution source on a concentration of a receptor point is calculated based on the pollutant concentration field distribution data, the sensitivity coefficient is substituted into a nonlinear response function including gas phase oxidation, liquid phase oxidation and heterogeneous reaction, and a pollutant superposition influence coefficient is calculated; A single-source transmission influence coefficient is calculated based on the pollutant superposition influence coefficient, the single-source transmission influence coefficient is input into a transfer closure algorithm for calculating a pollutant multi-source coupling transmission effect, a coupling transmission influence coefficient is obtained, and the single-source transmission influence coefficient and the coupling transmission influence coefficient are combined based on a time sequence sliding window to generate the pollution source correlation strength matrixes.

6. The method of claim 1, wherein, Optimal emission reduction proportions of the pollution sources are calculated based on the pollution source correlation strength matrixes and pollutant distribution weight coefficients, dynamic emission regulation and control parameters of the pollution sources are calculated according to the optimal emission reduction proportions, a multi-pollution source linkage control strategy is established based on the dynamic emission regulation and control parameters and the pollutant transmission path data, and coordinated regulation and control instructions of the pollution sources are generated, including: The pollution source correlation strength matrix and the pollution distribution weight coefficient are input into a nonlinear programming model, the nonlinear programming model includes an objective function of maximizing pollution reduction and minimizing economic cost, and the optimal emission reduction proportion of each pollution source is obtained by solving the nonlinear programming model through a genetic algorithm; A state space model including pollution source characteristics, meteorological conditions and environmental capacity is established according to the optimal emission reduction proportion, a Kalman filtering algorithm is used to estimate the system state in real time, and the state estimation result is input into a model predictive controller to calculate the dynamic emission regulation parameters of each pollution source; A pollution source weighted directed graph is constructed based on the dynamic emission regulation parameters and the pollution transmission path data, a community discovery algorithm is used to identify the strongly correlated pollution source groups in the weighted directed graph, and a differentiated group control strategy is generated according to the dynamic emission regulation parameters of each group to establish a multi-pollution source linkage control strategy; A hierarchical control structure is constructed based on the multi-pollution source linkage control strategy, a feedback compensation mechanism is set in the hierarchical control structure, actual pollution concentration change data is input into the feedback compensation mechanism for dynamic correction, and collaborative regulation instructions for each pollution source are generated.

7. The method of claim 6, wherein, A pollution source weighted directed graph is constructed based on the dynamic emission regulation parameters and the pollution transmission path data, a community discovery algorithm is used to identify the strongly correlated pollution source groups in the weighted directed graph, and a differentiated group control strategy is generated according to the dynamic emission regulation parameters of each group to establish a multi-pollution source linkage control strategy including: The dynamic emission regulation parameters are taken as node characteristic values, node connection relationships are constructed based on pollution transmission path data, edge characteristic values are obtained by updating the transmission amount between nodes in an exponential decay manner, and a pollution source weighted directed graph is constructed according to the node characteristic values and the edge characteristic values; The actual transmission strength and the network reference transmission strength between nodes are calculated based on the pollution source weighted directed graph, the actual transmission strength and the network reference transmission strength are input into the community discovery algorithm, and an initial group division scheme is obtained by iteratively optimizing the group division index; The transmission distance coefficient and the meteorological similarity coefficient between groups are calculated based on the initial group division scheme, the transmission distance coefficient and the meteorological similarity coefficient are combined with the group division index to construct a group optimization criterion, the initial group division scheme is optimized according to the group optimization criterion to obtain a target pollution source group, a group control model including pollution reduction amount and regulation cost is constructed for the target pollution source group, and a pollution source group regulation target is obtained by solving the group control model; The connection degree coefficient, the transmission flux coefficient and the position correlation coefficient of the nodes in each pollution source group are calculated, the connection degree coefficient, the transmission flux coefficient and the position correlation coefficient are combined to obtain a node priority coefficient, and the nodes in the pollution source group are graded based on the node priority coefficient to obtain a group internal grading scheme. According to the group internal grading scheme and the pollution source group regulation target, a group game model containing pollution reduction amount and regulation cost is constructed, a differentiated group control strategy is obtained by solving the group game model, a real-time monitoring data is collected to calculate a regulation correction amount, the regulation correction amount is input into a compensation model to adjust the differentiated group control strategy in real time, and a multi-pollution source linkage control strategy is established.

8. An electronic device, comprising: Comprise: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to invoke the instructions stored in the memory to perform the method of any one of claims 1 to 7.

9. A computer-readable storage medium having stored thereon computer program instructions, wherein, The computer program instructions, when executed by the processor, implement the method of any one of claims 1 to 7. The computer program instructions, when executed by the processor, implement the method of any one of claims 1 to 7.

Images