A method for tracing the source of VOCs generated locally by O3

Abstract

This invention discloses a method for tracing the source of VOCs in localized O3 generation, relating to the field of atmospheric environmental protection technology. The scheme includes: based on air quality monitoring and meteorological observation data, simulating hourly hydroxyl radical concentrations and in-situ O3 generation rates using an observation-based model, calculating VOCs photochemical reaction consumption, analyzing the industry sources of VOCs photochemical reaction consumption, identifying target industries with high contribution rates using machine learning, analyzing minute-level O3 monitoring concentrations and meteorological observation data through conditional variable probability functions to generate spatial heat maps of enterprises, and quickly identifying high-contributing target enterprises by combining these with the target industries. This allows for three-level source tracing at the industry, regional, and enterprise levels, achieving dynamic and accurate source tracing for localized O3 generation, providing refined management of VOCs emission sources in localized O3 generation, and providing a scientific basis for precise prevention and control of O3 generation.

Description

Technical Field

[0001] This application relates to the field of atmospheric environmental protection technology, and in particular to a method for tracing the source of VOCs generated locally by O3. Background Technology

[0002] Atmospheric ozone (O3) is not only a significant pollutant in assessing air quality, but high concentrations of O3 can also harm human health and inhibit vegetation growth. Volatile organic compounds (VOCs), as important precursors to O3, undergo photochemical reactions in the atmosphere, which are a key step in the formation of high concentrations of O3. Therefore, accurately identifying the sources of VOCs is crucial for controlling O3 formation.

[0003] Currently, my country's air pollution control has entered a stage of refined management. Accurately locating pollution sources at the local scale and implementing targeted VOCs control has become an urgent need to improve O3 control efficiency. To study the impact of VOCs emissions on O3 formation, existing source tracing methods involve extensive VOCs monitoring in different regions, combined with Lagrange particle diffusion models, emission source spectra, and machine learning to obtain VOCs emission characteristics from different pollution sources, thereby achieving source localization. However, due to the high reactivity of VOCs, up to 80% can be consumed through photochemical reactions during the process from the pollution source to the monitoring point. Therefore, accurately analyzing the VOCs consumed in this process (VOCs-C...) is crucial. OH The source of VOCs-C is a key prerequisite for reducing O3 formation. Currently, at the local scale... OH Current source tracing methods are mainly based on the Positive Matrix Factorization (PMF) model. However, their analytical results can only identify macro-level industry categories such as combustion sources and transportation sources, lacking the ability to quantify source categories that significantly contribute to local O3 generation. This makes it impossible to accurately locate specific pollution sources (e.g., a particular enterprise), and there is a lag, leading to a disconnect between the source tracing results and the actual needs for refined management. In summary, the analytical results of existing technologies are insufficient to meet the needs of refined management of local atmospheric O3 generation. Therefore, there is an urgent need to construct a dynamic source tracing technology that can connect "pollution source - VOCs - O3 generation". Summary of the Invention

[0004] This specification provides a method for tracing the source of VOCs generated locally by O3, in order to solve the problem that existing technologies cannot meet the requirements for refined control of VOCs emission sources.

[0005] To solve the above-mentioned technical problems, the embodiments in this specification are implemented as follows:

[0006] Firstly, the embodiments of this specification provide a method for tracing the source of VOCs generated locally by O3, including:

[0007] Based on air quality monitoring data and meteorological observation data, an observation-based model is used to obtain the hourly concentration of hydroxyl radicals and the in-situ O3 generation rate at the source to be traced. Based on the hydroxyl radical concentration and the reaction rate constants of each VOC species with hydroxyl radicals, the photochemical age method is used to calculate the VOCs photochemical reaction consumption, and the industry sources of the VOCs photochemical reaction consumption are analyzed. The VOCs photochemical reaction consumption refers to the amount of VOCs consumed during the transport process through photochemical reactions with hydroxyl radicals.

[0008] Using the in-situ O3 generation rate as the target variable, a machine learning model is used to identify target industries; the target industries represent industries that emit VOCs whose contribution to the O3 generation rate meets the first preset condition.

[0009] Based on minute-level O3 monitoring concentration and the meteorological observation data, the target area is identified hourly through a conditional variable probability function, and a spatial heat map of the enterprise is generated; the target area represents the region where the VOCs emission value of the conditional variable probability function meets the second preset condition.

[0010] The target industry is fused with the spatial heat map of the enterprise to identify the target enterprise; the target enterprise is characterized by emitting VOCs that meet the first preset condition and the second preset condition.

[0011] Secondly, the embodiments of this specification provide a VOCs tracing device for local O3 generation, comprising:

[0012] The determination module is used to input observation-based models based on air quality monitoring data and meteorological observation data to obtain the hourly concentration of hydroxyl radicals and the in-situ O3 generation rate of the local source to be traced. Based on the hydroxyl radical concentration and the reaction rate constants of each VOC species with hydroxyl radicals, the photochemical reaction consumption of VOCs is calculated by photochemical ageing, and the industry source of VOC photochemical reaction consumption is analyzed. The VOC photochemical reaction consumption refers to the amount of VOCs consumed during the photochemical reaction with hydroxyl radicals during transport.

[0013] The first identification module is used to identify target industries using the in-situ O3 generation rate as the target variable and through a machine learning model; the target industry represents the industry that emits VOCs and whose contribution to the O3 generation rate meets the first preset condition.

[0014] The second identification module is used to identify target areas hourly based on minute-level O3 monitoring concentration and the meteorological observation data, and generate a spatial heat map of the enterprise through a conditional variable probability function; the target area represents the area where the value of the conditional variable probability function meets the second preset condition for VOC emissions;

[0015] The third identification module is used to fuse the target industry with the enterprise spatial heat map to identify the target enterprise; the target enterprise represents an enterprise that emits VOCs and meets the first preset condition and the second preset condition.

[0016] One embodiment of this specification achieves the following beneficial effects: Based on air quality monitoring and meteorological observation data, an observation-based model simulates hourly hydroxyl radical concentration and in-situ O3 generation rate, calculates VOCs photochemical reaction consumption, analyzes the industry sources of VOCs photochemical reaction consumption, identifies target industries with high contribution through machine learning, analyzes minute-level O3 monitoring concentration and meteorological observation data through conditional variable probability function analysis, generates enterprise spatial heat maps, and can quickly identify target enterprises with high contribution by combining target industries. It can perform three-level source tracing at the industry, regional, and enterprise levels, realize dynamic and accurate source tracing for local O3 generation, provide refined management of VOCs emission sources of local O3 generation, and provide a scientific basis for precise prevention and control of O3 generation. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments or prior art of this specification, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating a VOCs source tracing method for local O3 generation provided in this specification embodiment;

[0019] Figure 2 This diagram illustrates an application scenario of a VOCs source tracing method for local O3 generation, as provided in the embodiments of this specification.

[0020] Figure 3 VOCs-C of each species output by the PMF model provided in the embodiments of this specification OH A diagram illustrating concentration and contribution;

[0021] Figure 4 A schematic diagram of SHAP values ​​for various industries provided in the embodiments of this specification;

[0022] Figure 5 A schematic diagram of the spatial distribution of CBPF values ​​for a certain hour provided in the embodiments of this specification;

[0023] Figure 6This is a schematic diagram of a VOCs source tracing device for local O3 generation, provided as an embodiment of this specification. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of one or more embodiments of this specification clearer, the technical solutions of one or more embodiments of this specification will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of them. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of one or more embodiments of this specification.

[0025] The technical solutions provided in the various embodiments of this specification are described in detail below with reference to the accompanying drawings.

[0026] The following is a detailed description of a VOCs source tracing method for local O3 generation, based on the embodiments provided in the specification, in conjunction with the accompanying drawings.

[0027] Figure 1 This is a flowchart illustrating a VOCs source tracing method for localized O3 generation provided in an embodiment of this specification. From a programming perspective, the executor of the process can be a program hosted on an application server or an application client. From a hardware perspective, the executor of the process can be a terminal device; this embodiment does not impose any particular limitation on this.

[0028] like Figure 1 As shown, the process may include the following steps:

[0029] Step 110: Based on air quality monitoring data and meteorological observation data, input the observation-based model to obtain the hourly concentration of hydroxyl radicals and the in-situ O3 generation rate of the local source to be traced; based on the hydroxyl radical concentration and the reaction rate constant of each VOC species with hydroxyl radicals, calculate the VOCs photochemical reaction consumption by photochemical ageing, and analyze the industry source of the VOCs photochemical reaction consumption; the VOCs photochemical reaction consumption refers to the amount of VOCs consumed during the transmission process through photochemical reactions with hydroxyl radicals.

[0030] In the embodiments of this specification, local air quality monitoring data and meteorological observation data to be traced are collected and input into an observation-based mechanism model (OBM). The model calculates and outputs hourly hydroxyl radical (·OH) concentration and in-situ O3 generation rate through a built-in photochemical reaction mechanism. The air quality monitoring data may include VOCs, NOx, CO, SO2, O3 concentrations, etc.; the meteorological observation data may include temperature, humidity, wind speed, wind direction, etc.

[0031] VOCs refer to organic compounds that are easily volatile at room temperature. Species can refer to specific organic compounds, such as toluene, ethylene, and propionaldehyde. Using photochemical aging, combined with hourly ·OH concentrations obtained from OBM simulations and known reaction rate constants between each VOC species and ·OH, the amount of VOCs consumed during transport through photochemical reactions with ·OH is calculated, i.e., the VOCs photochemical consumption (VOCs-C). OH The study considers the consumption of VOCs due to photochemical reactions during their journey from emission sources to monitoring points, which facilitates the development of emission reduction measures for industries that actually participate in O3 generation and emit VOCs.

[0032] Step 120: Using the in-situ O3 generation rate as the target variable, identify the target industry through a machine learning model; the target industry represents the industry that emits VOCs and whose contribution to the O3 generation rate meets the first preset condition.

[0033] In this embodiment, the in-situ O3 generation rate is used as the target variable, and a machine learning algorithm (such as Gradient Boosting Decision Tree X-GBoost) is used to identify the target industry. Industries are ranked from largest to smallest based on their contribution to O3 generation, with the first preset condition being the top two industries contributing the most to O3 generation.

[0034] Step 130: Based on minute-level O3 monitoring concentration and the meteorological observation data, the target area is identified hourly through the conditional variable probability function, and a spatial heat map of the enterprise is generated; the target area represents the area where the VOCs emission value of the conditional variable probability function meets the second preset condition.

[0035] In the embodiments of this specification, a conditional variable probability function (CBPF) model is constructed to evaluate the contribution probability of different regions to O3 generation. The second preset condition can be that the value of the conditional variable probability function is greater than or equal to 0.5.

[0036] By combining the target area with map information, a spatial heat map of enterprises can be generated, which can intuitively show the distribution of enterprises emitting VOCs within the target area.

[0037] Step 140: Merge the target industry with the enterprise spatial heat map to identify the target enterprise; the target enterprise represents an enterprise that emits VOCs and meets the first preset condition and the second preset condition.

[0038] In the embodiments of this specification, through spatial analysis and industry matching, enterprises that emit VOCs and simultaneously meet the first preset condition (industries with high contribution to O3 generation rate) and the second preset condition (located in high probability contribution areas) are identified as target enterprises.

[0039] It should be understood that the order of some steps in the methods described in one or more embodiments of this specification may be interchanged according to actual needs, or some steps may be omitted or deleted.

[0040] In the embodiments of this specification, based on air quality monitoring and meteorological observation data, an observation-based model is used to simulate hourly hydroxyl radical concentration and in-situ O3 generation rate, calculate VOCs photochemical reaction consumption, analyze the industry sources of VOCs photochemical reaction consumption, and identify target industries with high contribution through machine learning. By analyzing minute-level O3 monitoring concentration and meteorological observation data through conditional variable probability functions, a spatial heat map of enterprises is generated. Combined with the target industry, target enterprises with high contribution can be quickly identified. Three-level source tracing can be carried out at the industry, regional, and enterprise levels, realizing dynamic and accurate source tracing for local O3 generation, providing refined management of VOCs emission sources of local O3 generation, and providing a scientific basis for precise prevention and control of O3 generation.

[0041] based on Figure 1 In addition to the method described in the embodiments of this specification, some specific implementation schemes of the method are also provided, which will be described below.

[0042] Optionally, the method of using the in-situ O3 generation rate as the target variable and identifying the target industry through a machine learning model, as described in the embodiments of this specification, may specifically include:

[0043] The VOCs photochemical reaction consumption is input into a positive definite matrix factorization model, and combined with VOCs emission industry characteristic tracers, the specific industry for VOCs photochemical reaction consumption is determined.

[0044] Using the in-situ O3 generation rate as the target variable, the specific industry of VOCs photochemical reaction consumption is input into the machine learning model, and combined with Shapley additive interpretation, the target industry is obtained.

[0045] In the embodiments of this specification, the calculated VOCs photochemical reaction consumption (VOCs-C) will be used. OH The data is input into a positive definite matrix factorization (PMF) model. Simultaneously, through PMF model analysis and combined with known VOCs emission industry characteristic tracers, the specific industry sources of VOCs photochemical reaction consumption are determined. These characteristic tracers help distinguish the VOCs characteristics emitted by different industries.

[0046] Using the in-situ O3 generation rate as the target variable, industry-specific data on VOCs photochemical reaction consumption obtained in the previous step are used as input features and fed into a machine learning model (such as X-GBoost). The Shapley Additive Interpretation (SHAP) method is used to interpret the prediction results of the machine learning model, obtaining the contribution of each industry to the O3 generation rate (SHAP value). The SHAP value indicates the level of contribution of each industry to local O3 generation. Based on the magnitude of the SHAP value, key VOCs industries that contribute significantly to local O3 generation are dynamically quantified and identified, i.e., the target industries.

[0047] Optionally, the formula for calculating the VOCs photochemical reaction consumption in the embodiments of this specification can be [VOCs-C OH ] i,j =[M-VOCs] i,j ×(exp(K i ×[OH] j ×Δt j )-1), where, [VOCs-C OH ] i,j Let [M-VOCs] be the concentration of VOC species i lost through photochemical reaction at time j. i,j K represents the monitoring concentration of VOC species i at time j. i Let be the rate constant for the reaction between VOC species i and hydroxyl radicals, and [OH] be the hydroxyl radical. j Let Δt be the concentration of hydroxyl radicals at time j. j Let j be the photochemical age at time j.

[0048] In the embodiments of this specification, the maximum value of the ratio of the monitored concentrations of para-xylene (X) to ethylbenzene (E) in the early morning of the target day is statistically analyzed, and the photochemical age is calculated based on the X / E concentration ratio method.

[0049] Specifically, the monitored concentrations of m- / p-xylene (X) and ethylbenzene (E) were obtained, and the maximum value of the ratio of the monitored concentrations of X to E (X / E) at dawn was recorded as the initial emission ratio.

[0050] The formula for calculating Δt based on the X / E concentration ratio method is as follows:

[0051]

[0052] Where Δt is the photochemical age, in seconds; [OH] is the average concentration of OH on the target day, in molecules per centimeter. -3 ;K x Let X be the rate constant for the reaction between X and ·OH, 18.9 × 10⁻⁶. -12 cm 3 ·molecule -1·s -1 ;K E Let be the rate constant for the reaction between E and ·OH, 7.0 × 10⁻⁶. -12 cm 3 ·molecule -1 ·s -1 ; The initial emission ratio for the day is dimensionless. Let t be the ratio of the monitored concentrations of X to E, which is dimensionless.

[0053] Optionally, the method described in the embodiments of this specification for identifying target areas hourly and generating enterprise spatial heat maps based on minute-level O3 monitoring concentrations and meteorological observation data through a conditional variable probability function may specifically include:

[0054] Based on minute-level O3 monitoring concentration and the aforementioned meteorological observation data, an O3 concentration threshold is set. Combining wind speed and direction, the probability function of the conditional variable is calculated. The formula for calculating the probability function of the conditional variable is as follows: Where, mθ i,j nθ represents the number of times the concentration of factor j exceeds the threshold under wind direction i. i,j The number of times factor j contributes to the concentration of wind direction i;

[0055] The numerical values ​​of the conditional variable probability function are overlaid with geographic information to generate a spatial heat map of the enterprise.

[0056] In the embodiments described in this specification, high temporal resolution O3 monitoring concentration data are collected from the local area to be traced. Simultaneously, corresponding meteorological observation data, particularly wind speed and direction data, are collected. Wind speed and direction are key factors determining the transport path and diffusion range of pollutants.

[0057] In practice, the minute-level O3 monitoring concentration and the meteorological observation data can be preprocessed to ensure the time synchronization of the O3 monitoring concentration data and the meteorological observation data, and necessary data cleaning and outlier handling can be performed to ensure the accuracy of the analysis.

[0058] Based on local O3 pollution levels, air quality standards, or research objectives, a reasonable O3 concentration threshold should be set, such as 75th percentile of the monitored O3 concentration. This threshold is used to determine whether the O3 concentration exceeds the normal range, thereby identifying potentially polluted areas.

[0059] The conditional probability function (CBPF) is a statistical indicator used to assess the probability that a certain factor (such as O3 concentration) exceeds a set threshold under specific wind conditions.

[0060] For each wind direction range, the number of times the O3 monitoring concentration exceeded the threshold and the total number of observations under each wind direction condition were counted. These results were then substituted into the CBPF formula for calculation. The numerical calculation formula for the conditional variable probability function is as follows: Where, mθ i,j nθ represents the number of times the concentration of factor j exceeds the threshold under wind direction i. i,j The number of times factor j contributes to the concentration under wind direction i. Using the calculated CBPF values, the probability of O3 generation contribution in each region under different wind directions is evaluated hourly. Based on a second preset condition (e.g., CBPF value exceeding 0.5), regions that significantly contribute to O3 generation under the current meteorological conditions are identified as target regions.

[0061] The calculated CBPF values ​​are overlaid with map data from a Geographic Information System (GIS) to determine the specific geographic location of the target area. Combined with information on the distribution of businesses in the surrounding area (such as business location and industry type), businesses located within the target area are selected. Using GIS technology, the selected businesses are visualized according to their contribution probability to O3 generation (i.e., CBPF values), generating a spatial heatmap of the businesses. In the heatmap, different colors or icon sizes represent the degree of contribution of each business to O3 generation, thus intuitively displaying the spatial distribution of polluting businesses.

[0062] Based on minute-level O3 monitoring concentration and meteorological observation data, combined with the conditional variable probability function (CBPF) method, it is possible to effectively identify target areas that significantly contribute to O3 generation on an hourly basis and generate intuitive spatial heat maps of enterprises, providing strong support for the precise control of local atmospheric O3 generation.

[0063] Further, optionally, before inputting the VOCs photochemical reaction consumption into the positive definite matrix factorization model as described in the embodiments of this specification, the method may include:

[0064] Obtain the VOCs monitoring concentration and instrument detection limit at air quality monitoring stations, and determine the uncertainty of the VOCs consumption by photochemical reaction;

[0065] The uncertainty in determining the amount of VOCs consumed through photochemical reaction specifically includes:

[0066] If the monitored concentration of VOCs is lower than the detection limit of the instrument, the uncertainty is calculated as follows: unc = 4 × con, where unc is the uncertainty of the amount of VOCs consumed by the photochemical reaction, and con is the concentration of VOCs lost by the photochemical reaction.

[0067] If the VOCs monitoring concentration is higher than the instrument detection limit, the uncertainty is calculated as follows: unc = 0.1 × con.

[0068] In the embodiments described in this specification, VOCs monitoring concentration data are obtained from air quality monitoring stations. The monitoring concentration data is measured using high-precision instruments such as gas chromatography-mass spectrometry (GC-MS). Simultaneously, the detection limit of the instrument used needs to be determined. The detection limit refers to the lowest concentration that the instrument can reliably detect; for measurements below this limit, the accuracy and reliability will be affected.

[0069] Uncertainty is an important indicator for measuring the reliability of measurement results. When determining the uncertainty of VOCs consumption through photochemical reactions, different calculation formulas are needed based on the relationship between the monitored VOCs concentration and the instrument's detection limit.

[0070] If the VOCs monitoring concentration is lower than the instrument detection limit, the calculation formula is: unc = 4 × con. When the VOCs monitoring concentration is lower than the instrument detection limit, the measurement result has a large uncertainty.

[0071] If the VOCs monitoring concentration is higher than the instrument detection limit, the calculation formula is: unc=0.1×con. When the VOCs monitoring concentration is higher than the instrument detection limit, the reliability of the measurement results is relatively high.

[0072] The photochemical reaction consumption of VOCs and the corresponding uncertainty are input into a positive definite matrix factorization model. A signal-to-noise ratio (SNR) threshold is set, and species with SNRs below the threshold are eliminated. The robust solution with convergent residuals and stable Q / Qexp values ​​is selected as the final output. Combined with the PMF model for analysis, industries that contribute significantly to O3 generation of VOCs emissions can be identified more accurately.

[0073] Figure 2 This diagram illustrates an application scenario of a VOCs source tracing method for local O3 generation, as provided in the embodiments of this specification.

[0074] To facilitate understanding, examples will be used, such as... Figure 2 As shown, based on the invented source tracing method, taking an industrial park in a certain city as the research area, a certain O3 polluted day in June 2023 as the research period, and a certain central monitoring station as the target monitoring station, a case study source tracing was conducted as follows:

[0075] Step 201: Obtain hourly monitoring data of VOCs, NO, NO2, CO, SO2, and O3 concentrations, as well as observation data of temperature, relative humidity, wind speed, and wind direction, and minute-level monitoring data of O3 concentration, wind speed, and wind direction, and instrument detection limits from the target central station on a polluted day in June 2023; obtain information on pollution sources around the target central station (company name, longitude, latitude, industry type, main production line, main products, etc.) and characteristic tracer species of different VOCs-emitting industries; based on literature review, obtain the reaction rate constants of each VOCs species with ·OH.

[0076] Step 202: Based on the hourly monitoring concentrations of VOCs, NO, NO2, CO, SO2, and O3, as well as the temperature and humidity data obtained in Step 201, missing values ​​are screened and supplemented using interpolation. The units of the pollutant monitoring concentrations and meteorological observation data are converted so that the units of VOCs, NO, NO2, SO2, and O3 are ppb, CO is ppm, temperature is K, and relative humidity is %. The OBM model is run with VOCs, NO, NO2, SO2, O3, CO, temperature, and relative humidity as input variables. The output of the OBM model is post-processed to obtain the hourly ·OH concentration and O3 generation rate.

[0077] Based on the acquired monitoring concentrations of X and E, the maximum ratio of the monitoring concentrations of X to E during the early morning (1:00-5:00) is calculated as the initial emission ratio for the day; the photochemical age Δt is calculated based on the X / E concentration ratio method, as shown in the following formula:

[0078]

[0079] Step 203: Based on the 0-23 hour VOCs monitoring concentrations obtained in Step 201, the reaction rate constants of each VOCs species with ·OH, the hourly ·OH concentrations obtained in Step 202, and the hourly photochemical age Δt, the hourly VOCs-C is calculated using the photochemical age method. OH The formula is as follows:

[0080] VOCs-C OH ] i,j =[M-VOCs] i,j ×(exp(K i ×[OH] j ×Δt j )-1)

[0081] Step 204: Based on the instrument detection limit obtained in Step 201, screen the sample IDs with VOCs monitoring concentrations below the detection limit, and calculate the VOCs-C concentration for that batch. OH The uncertainty is calculated using the following formula:

[0082] unc = 4 × con

[0083] Select sample IDs with VOCs monitoring concentrations higher than the detection limit, and calculate the VOCs-C of that batch. OH The uncertainty is calculated using the following formula:

[0084] unc = 0.1 × con

[0085] Hourly VOCs-C of each species OHInput the concentration and uncertainty data files into the PMF model to analyze VOCs-C OH For specific industries, species with a signal-to-noise ratio below 0.5 are excluded, and robust solutions with convergent residuals and stable Q / Qexp values ​​are selected as the final output.

[0086] Figure 3 VOCs-C of each species output by the PMF model provided in the embodiments of this specification OH A diagram illustrating concentration and contribution.

[0087] like Figure 3 As shown, based on the output of VOCs-C for each species OH Based on the concentration and contribution rate, combined with the tracer information of different industry characteristics and pollution sources around the site obtained in step 201, the industry type was determined to be a mixed industrial source - solvent use / petrochemical industry and industrial emissions - machinery manufacturing industry.

[0088] Step 205: Based on the hourly industry concentration contributions, the hourly NO2 and NO monitoring concentrations obtained in Step 201, and the observed data of temperature and relative humidity as input variables, and using the O3 generation rate obtained in Step 202 as the target variable, train the gradient boosting decision tree (X-GBoost) model, evaluate the prediction results of different model parameters, and output the optimal parameter combination and the corresponding model performance index. The R-squared value is selected as the model evaluation index. 2 RMSE, MMAE, correlation coefficient R 2 The calculation formula is as follows:

[0089]

[0090] Where n is the total sample size, which is dimensionless; y i For the i-th actual value, ppb·h -1 ; For the j-th predicted value, ppb·h -1 ; The average of all actual values, ppb·h -1 .

[0091] The formula for calculating the root mean square error (RMSE) is as follows:

[0092]

[0093] The formula for calculating the Mean Absolute Error (MAE) is as follows:

[0094]

[0095] Model parameters are set based on the optimal parameter combination corresponding to the output indicators. Interpretive analysis is performed using the SHAP method to obtain and sort the SHAP values ​​for each industry.

[0096] Figure 4 This is a schematic diagram of SHAP values ​​for various industries provided in the embodiments of this specification.

[0097] like Figure 4 As shown, industries with high hourly SHAP values ​​are identified through dynamic quantification, thus pinpointing target industries that significantly contribute to local O3 generation and VOC emissions. The hourly SHAP values ​​for each industry are shown in Table 1.

[0098] Table 1

[0099]

[0100] Step 206: Based on the minute-level O3 monitoring concentration, wind speed, and wind direction observation data obtained in Step 201, using the 75th percentile of the O3 monitoring concentration as a threshold, and combining wind speed and wind direction, construct a CBPF model. The calculation formula for CBPF values ​​in different wind directions is as follows:

[0101]

[0102] Figure 5 This is a schematic diagram of the spatial distribution of CBPF values ​​for a certain hour, provided as an embodiment of this specification.

[0103] like Figure 5 As shown, the CBPF values ​​and corresponding wind speeds for each wind direction at a certain hour on a polluted day are spatially overlaid with the map of the area surrounding the site to identify target areas with VOC emissions that have a high probability of contributing to O3 generation.

[0104] Step 207: Combining the surrounding pollution source information obtained in Step 201, identify the spatial distribution of enterprises that have a high probability of contributing to O3 generation hourly, and obtain enterprise spatial heat map information.

[0105] Step 208: Based on the acquired enterprise spatial heat map information, and combined with the industries that contribute significantly to local O3 obtained in Step 205, identify target enterprises that emit VOCs and contribute significantly to local O3 generation hourly, thereby achieving dynamic tracing of these enterprises. Table 2 shows the VOCs-emitting enterprises that contribute significantly to O3 generation on a specific hour of a polluted day.

[0106] Table 2

[0107] Company Name latitude and longitude A Packaging Co., Ltd. 108.6**，34.11** B Printing Co., Ltd. 108.6**，34.10** C Packaging Industry Co., Ltd. 108.64**，34.11** D Pharmaceutical Packaging Co., Ltd. 108.65**，34.12** E-Materials Technology Co., Ltd. 108.67**，34.11** F Plastics Co., Ltd. 108.67**，34.10** G Materials Development Co., Ltd. 108.66**，34.10** H Pharmaceutical Co., Ltd. 108.38**，34.06**

[0108] This invention, based on air quality monitoring and meteorological observation data, simulates ·OH concentration and in-situ O3 formation rate to calculate VOCs-C. OH And analyze VOCs-C OHThe invention identifies key VOCs industries that significantly contribute to local O3 generation by combining machine learning algorithms with hourly quantitative analysis. Simultaneously, it constructs a conditional variable probability function based on minute-level O3 monitoring concentrations to hourly identify spatial heatmap information of enterprises with a high probability of contributing to O3 generation. Finally, by overlaying the spatial heatmap information of key VOCs industries and enterprises, the invention dynamically traces VOCs-emitting enterprises that significantly contribute to local O3 generation. This invention achieves dynamic and accurate source tracing for local O3 generation, providing technical support for the refined management of VOCs in local atmospheric O3 generation.

[0109] Figure 6 This is a schematic diagram of a VOCs source tracing device for local O3 generation, provided as an embodiment of this specification.

[0110] Corresponding to the method embodiment, this embodiment also provides a VOCs source tracing device for local O3 generation, which may include:

[0111] The determination module 602 is used to input observation-based models based on air quality monitoring data and meteorological observation data to obtain the hourly concentration of hydroxyl radicals and the in-situ O3 generation rate of the local source to be traced; based on the hydroxyl radical concentration and the reaction rate constants of each VOC species with hydroxyl radicals, the photochemical reaction consumption of VOCs is calculated by photochemical ageing, and the industry source of the VOCs photochemical reaction consumption is analyzed; the VOCs photochemical reaction consumption refers to the amount of VOCs consumed during the transport process through photochemical reactions with hydroxyl radicals;

[0112] The first identification module 604 is used to identify target industries using the in-situ O3 generation rate as the target variable and through a machine learning model; the target industry represents the industry that emits VOCs whose contribution to the O3 generation rate meets the first preset condition.

[0113] The second identification module 606 is used to identify target areas hourly based on minute-level O3 monitoring concentration and the meteorological observation data, and generate a spatial heat map of the enterprise through a conditional variable probability function; the target area represents the area where the value of the conditional variable probability function meets the second preset condition for emitting VOCs.

[0114] The third identification module 608 is used to fuse the target industry with the enterprise spatial heat map to identify the target enterprise; the target enterprise represents an enterprise that emits VOCs and meets the first preset condition and the second preset condition.

[0115] Optionally, the method of using the in-situ O3 generation rate as the target variable and identifying the target industry through a machine learning model, as described in the embodiments of this specification, may specifically include:

[0116] The VOCs photochemical reaction consumption is input into a positive definite matrix factorization model, and combined with VOCs emission industry characteristic tracers, the specific industry for VOCs photochemical reaction consumption is determined.

[0117] Using the in-situ O3 generation rate as the target variable, the specific industry of VOCs photochemical reaction consumption is input into the machine learning model, and combined with Shapley additive interpretation, the target industry is obtained.

[0118] Optionally, the method described in the embodiments of this specification for identifying target areas hourly and generating enterprise spatial heat maps based on minute-level O3 monitoring concentrations and meteorological observation data through a conditional variable probability function may specifically include:

[0119] Based on minute-level O3 monitoring concentration and the aforementioned meteorological observation data, an O3 concentration threshold is set. Combining wind speed and direction, the probability function of the conditional variable is calculated. The formula for calculating the probability function of the conditional variable is as follows: Where, mθ i,j nθ represents the number of times the concentration of factor j exceeds the threshold under wind direction i. i,j The number of times factor j contributes to the concentration of wind direction i;

[0120] The numerical values ​​of the conditional variable probability function are overlaid with geographic information to generate a spatial heat map of the enterprise.

[0121] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0122] The foregoing has described specific embodiments of this specification; other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims may be performed in a different order than those shown in the embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily have to follow the specific or sequential order shown to achieve the desired result. The various embodiments in this specification are described in a progressive manner; similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments.

[0123] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0124] The above description is merely an embodiment of this specification and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of this application should be included within the scope of the claims of this application.

Claims

1. A method for tracing the sources of VOCs generated locally by O3, characterized in that, include: Based on air quality monitoring data and meteorological observation data, an observation-based model is used to obtain the hourly concentration of hydroxyl radicals and the in-situ O3 generation rate at the source to be traced. Based on the hydroxyl radical concentration and the reaction rate constants of each VOC species with hydroxyl radicals, the photochemical age method is used to calculate the VOCs photochemical reaction consumption, and the industry sources of the VOCs photochemical reaction consumption are analyzed. The VOCs photochemical reaction consumption refers to the amount of VOCs consumed during the transport process through photochemical reactions with hydroxyl radicals. Using the in-situ O3 generation rate as the target variable, a machine learning model is used to identify target industries; the target industries represent industries that emit VOCs whose contribution to the O3 generation rate meets the first preset condition. Based on minute-level O3 monitoring concentration and the meteorological observation data, the target area is identified hourly through a conditional variable probability function, and a spatial heat map of the enterprise is generated; the target area represents the region where the VOCs emission value of the conditional variable probability function meets the second preset condition. The target industry is fused with the enterprise spatial heat map to identify the target enterprise; the target enterprise represents an enterprise that emits VOCs and meets the first preset condition and the second preset condition. The step of using the in-situ O3 generation rate as the target variable and identifying the target industry through a machine learning model specifically includes: The VOCs photochemical reaction consumption is input into a positive definite matrix factorization model, and combined with VOCs emission industry characteristic tracers, the specific industry for VOCs photochemical reaction consumption is determined. Using the in-situ O3 generation rate as the target variable, the specific industry of VOCs photochemical reaction consumption is input into the machine learning model, and combined with Shapley additive interpretation, the target industry is obtained.

2. The method according to claim 1, characterized in that, The formula for calculating the VOCs photochemical reaction consumption is as follows: ,in, VOCs species exist The concentration lost due to photochemical reactions occurring constantly. VOCs species exist Real-time monitoring of concentration, VOCs species The rate constant of the reaction with hydroxyl radicals for The concentration of hydroxyl radicals at any given time for Time-based photochemical age.

3. The method according to claim 1, characterized in that, The minute-level By monitoring concentrations and the aforementioned meteorological observation data, and using a conditional variable probability function, the target area is identified hourly, and a spatial heat map of the enterprise is generated, specifically including: Based on minute level Monitor concentration and the aforementioned meteorological observation data, and set up The concentration threshold, combined with wind speed and direction, is used to calculate the numerical value of the conditional variable probability function. The formula for calculating the numerical value of the conditional variable probability function is as follows: ,in, Wind direction Lower factor The number of times the contribution concentration exceeds the threshold. Wind direction Lower factor The number of times the contribution concentration appears; The numerical values ​​of the conditional variable probability function are overlaid with geographic information to generate a spatial heat map of the enterprise.

4. The method according to claim 1, characterized in that, Before inputting the VOCs photochemical reaction consumption into the positive definite matrix factorization model, the method includes: Obtain the VOCs monitoring concentration and instrument detection limit at air quality monitoring stations, and determine the uncertainty of the VOCs consumption by photochemical reaction; The uncertainty in determining the amount of VOCs consumed through photochemical reaction specifically includes: If the monitored VOCs concentration is lower than the instrument's detection limit, then the uncertainty is calculated using the following formula: ,in, The uncertainty of the amount of VOCs consumed in the photochemical reaction. The concentration of VOCs lost through photochemical reactions; If the monitored VOCs concentration is higher than the instrument's detection limit, then the uncertainty is calculated using the following formula: .

5. A local-oriented approach The generated VOCs traceability device is characterized in that, include: The determination module is used to input observation-based models based on air quality monitoring data and meteorological observation data to obtain the hourly concentration and in-situ concentration of hydroxyl radicals at the source of the problem. Generation rate; based on the concentration of hydroxyl radicals and the reaction rate constants of each VOC species with hydroxyl radicals, the photochemical reaction consumption of VOCs is calculated by photochemical ageing, and the industry sources of the VOCs photochemical reaction consumption are analyzed; the VOCs photochemical reaction consumption refers to the amount of VOCs consumed during the transport process through photochemical reactions with hydroxyl radicals; The first identification module is used to identify the in-situ... The generation rate is used as the target variable, and a machine learning model is used to identify the target industry; the target industry representation is... Industries whose VOC emissions meet the first preset condition in terms of generation rate contribution; The second identification module is used for minute-level identification. By monitoring concentrations and meteorological observation data, and using a conditional variable probability function, target areas are identified hourly, and a spatial heat map of the enterprise is generated; the target area represents the region where the value of the conditional variable probability function satisfies a second preset condition for VOC emissions. The third identification module is used to fuse the target industry with the enterprise spatial heat map to identify the target enterprise; The target enterprise is defined as an enterprise that emits VOCs and meets the first preset condition and the second preset condition; The in-situ With generation rate as the target variable, a machine learning model is used to identify target industries, specifically including: The VOCs photochemical reaction consumption is input into a positive definite matrix factorization model, and combined with VOCs emission industry characteristic tracers, the specific industry for VOCs photochemical reaction consumption is determined. In the in-situ The generation rate is used as the target variable. The specific industry of VOCs photochemical reaction consumption is input into the machine learning model, and the target industry is obtained by combining Shapley additive interpretation.

6. The apparatus according to claim 5, characterized in that, The minute-level By monitoring concentrations and the aforementioned meteorological observation data, and using a conditional variable probability function, the target area is identified hourly, and a spatial heat map of the enterprise is generated, specifically including: Based on minute level Monitor concentration and the aforementioned meteorological observation data, and set up The concentration threshold, combined with wind speed and direction, is used to calculate the numerical value of the conditional variable probability function. The formula for calculating the numerical value of the conditional variable probability function is as follows: ,in, Wind direction Lower factor The number of times the contribution concentration exceeds the threshold. Wind direction Lower factor The number of times the contribution concentration appears; The numerical values ​​of the conditional variable probability function are overlaid with geographic information to generate a spatial heat map of the enterprise.

Images