An atmospheric environment ozone concentration prediction system based on chemical reaction mechanism

Abstract

The application discloses a kind of atmospheric environment ozone concentration prediction system based on chemical reaction mechanism, including obtaining weather forecast parameter, historical observation data cleaning analysis, analysis screening suitable model, model training and prediction, prediction result analysis, data visualization and so on functional module composition.This system is based on Lagrangian coordinate system, establishes the response relationship between meteorological parameter, pollutant concentration and ozone concentration, using thermodynamic equilibrium principle, chemical reaction equilibrium principle and photochemical reaction principle and other basic chemical reaction engineering theory, deduces the non-linear model of meteorological parameter, pollutant concentration and other parameters on ozone concentration influence, by the independent design of each software function module, with low energy consumption, low carbon emission, low operating cost, data automatic processing, easy to deploy, can be connected ecological cloud platform, system expansibility is strong, maintenance is easy, and computing efficiency is high.Therefore, it has broad market application prospect and good social and economic benefits.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of atmospheric environmental pollution prediction, and particularly relates to an ozone prediction system based on chemical reaction mechanism. BACKGROUND

[0002] Accurate prediction of ozone (O3) concentration has always been a relatively difficult problem to achieve, and prediction often needs to rely on large computers, and only after a large number of simulation calculations based on complex weather models can relatively accurate results be obtained. Generally speaking, this traditional prediction method is more suitable for areas with serious pollution, and may not be in line with the actual situation of areas with lighter air pollution. The hysteresis of the decrease in ozone concentration is often difficult to reflect through traditional models.

[0003] When the traditional model has problems, it is a very meaningful work to reselect a scientific and reasonable prediction path to replace the traditional model with high energy consumption and high cost and reduce carbon emissions. SUMMARY

[0004] The purpose of the present application is to construct an ozone prediction model with lower calculation amount, which is suitable for high, medium and low atmospheric environmental pollution conditions, to reduce the calculation amount of the prediction process and to reduce the calculation scheme of carbon emissions.

[0005] The purpose of the present application is achieved by an atmospheric environmental ozone concentration prediction system based on chemical reaction mechanism, characterized by comprising:

[0006] 1) System software functions, including obtaining weather forecast parameters, cleaning and analyzing historical observation data, analyzing and screening suitable models, model training and prediction, prediction result analysis, and data visualization;

[0007] 2) The software function modules of the system are independently designed, and can be run independently or in parallel for accelerated calculation, and can be processed by a single machine or easily deployed for cluster computing and cloud computing;

[0008] 3) Based on the Lagrangian coordinate system, the response relationship between meteorological parameters, pollutant concentration and ozone concentration is established;

[0009] 4) The non-linear model of the influence of meteorological parameters and pollutant concentration parameters on ozone concentration is derived based on the theories of thermodynamic equilibrium principle, chemical reaction equilibrium principle and photochemical reaction principle in basic chemical reaction engineering.

[0010] The ozone concentration prediction model can represent the influence of humidity W and air pressure P coupling on ozone concentration , which satisfies the relationship , wherein: k', P * , and b' are all fitting coefficient terms.

[0011] The ozone concentration prediction model described above characterizes the effect of the coupling of air temperature T and air pressure P on ozone concentration. The impact, satisfying The relationship is given by the formula, where a, b, c, and d all refer to the fitting coefficients; or the relationship can be obtained by removing... The simplified form of any one or two of the three terms in lnP.

[0012] The ozone concentration prediction model described above characterizes the effect of solar radiation intensity I on ozone concentration. The impact, satisfying The relationship is given by the formula: I0 refers to the intensity of the incident light; C refers to photochemical efficiency; C refers to the concentration of the light-absorbing substance (mol / L); t refers to the photochemical reaction time.

[0013] The ozone concentration prediction model described above is used to characterize the concentration of volatile organic compounds (VOCs). Nitrogen oxides inorganic concentration With ozone concentration Relationship, satisfying The relationship is given by the formula: α i δ k ∈ both refer to the fitting coefficients; here the relationship between VOCs and ozone applies to all types of organic compounds, including BVOCs, OVOCs, and organohalogenated compounds.

[0014] The ozone concentration prediction model described above is used to characterize particulate matter concentration C. PM With ozone concentration Relationship, satisfying The relationship is given by the formula, where α′ and β′ are the mean fitting coefficients.

[0015] Furthermore, by randomly selecting any combination of characteristic relationships, a relatively accurate ozone concentration prediction model can be obtained, especially with the combination of meteorological parameters.

[0016] This invention reveals the mechanistic relationship between meteorological parameters, particulate matter, VOCs, NOx inorganic matter and ozone concentration.

[0017] The fine particulate matter (PM) derived above in this invention 2.5 The prediction model for secondary atmospheric pollutants (PM1) also conforms to the features described in claim 1 and possesses any of the features described above.

[0018] Specifically, the technical solution of the present invention is as follows:

[0019] The atmospheric environment ozone concentration prediction model is composed of the following function modules: acquiring weather forecast parameters, cleaning and analyzing historical observation data, analyzing and screening suitable models, model training and prediction, prediction result analysis, data visualization, and data docking with the ecological cloud platform.

[0020] ① All models are based on the Lagrangian coordinate system;

[0021] ② Assuming that the gas micro-cluster reaches a pseudo-equilibrium state during the transmission process, the ozone concentration and the concentrations of VOCs and NOx and other substances satisfy the relationship Further combined with catalytic theory, the ozone concentration satisfies the relationship In addition, it can also be concluded that the relationship between ozone concentration and air pressure may satisfy The relationship with the reaction equilibrium constant satisfies

[0022] ③ Further combined with the principles of thermodynamics, the coupling effect of humidity and pressure on ozone concentration can be further deduced The coupling effect of temperature and pressure on ozone concentration can also be further deduced

[0023] ④ Combined with the Beer-Lambert law and photochemical basic theory such as light quantum yield, the relationship between ozone concentration and light intensity can be deduced

[0024] At the same time, by reanalyzing the atmospheric chemical process in this way, the causes and mechanisms of PM 2.5 and O3 pollution can be deconstructed from another aspect.

[0025] The advantages of the present application are:

[0026] ① Compared with the traditional atmospheric pollution prediction model, regional data of pressure field, temperature field, flow field and other meteorological parameters are not required, and local climate conditions do not need to be considered additionally;

[0027] ② Compared with the traditional atmospheric pollution prediction model, the data calibration accuracy requirement is not high, and the observation error has strong fault tolerance capability, so that the observation error of historical parameters will not affect the field data, thereby reducing the accuracy of the results;

[0028] ③ Compared with the traditional atmospheric pollution prediction model, the ozone concentration prediction is suitable for sites with complete historical observation data, and the ozone concentration prediction results can be obtained more accurately by supplementing the data through acquiring similar meteorological observation sites for sites with part of missing meteorological parameters;

[0029] (4) Compared with the traditional atmospheric pollution prediction model, the pollution source is hidden in the historical data and the wind direction data characteristics, and becomes a part of the parameter set through data training, without the need to additionally develop a source list;

[0030] (5) Similar to the offline prediction mode of the traditional atmospheric pollution prediction model, but without the need to establish a meteorological prediction model by itself, only meteorological parameters are obtained from the network online at the time of prediction, which are directly used for prediction;

[0031] (6) Strong scalability, which can adjust the model according to the number of parameters of historical observation data, but the number of parameters has a certain influence on the prediction accuracy of the model.

[0032] Therefore, this type of atmospheric environment ozone concentration prediction model has broad application prospects and good social and economic benefits. Performance includes:

[0033] (1) Low energy consumption, low carbon emission and low operation cost. Compared with the traditional atmospheric pollution prediction model, a large amount of computing resources can be saved, the cost of prediction can be reduced, and the power consumption generated in the calculation process can be reduced, which contributes to reducing carbon emissions under the condition of the same prediction accuracy.

[0034] (2) Data automatic processing, easy deployment and can be connected to an ecological cloud platform. The functions of software modules such as obtaining weather forecast parameters, cleaning and analyzing historical observation data, analyzing and screening suitable models, model training and prediction, prediction result analysis, data visualization and data docking with an ecological cloud platform can be automatically run. The software function modules of the system are independently designed, which can be run alone or in parallel to accelerate the calculation, and can be processed by a single machine or easily deployed for cluster computing and cloud computing.

[0035] (3) Strong system scalability and easy maintenance. The core algorithm model can be replaced or optimized and upgraded as needed.

[0036] (4) High calculation efficiency. Since the meteorological model does not need to be run by itself, a large amount of computing resources can be saved, and the optimization and upgrading of the core algorithm can be concentrated. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 is a comparative analysis diagram of the influence of the ozone concentration of the present application on the coupling effect of humidity and pressure.

[0038] Figure 2 is an analysis diagram of the influence of the ozone concentration of the present application on the coupling effect of temperature and pressure.

[0039] Figure 3 is a comparison diagram of the prediction results of the ozone concentration of the present application using formulas (4), (8), (16), (20) and (23) and the actual values.

[0040] Figure 4 is the time sequence diagram of the prediction results and actual values of the ozone concentration of the application using the formula (4), (8), (16), (20), (23) derivation model.

[0041] Figure 5 is the system software architecture design schematic diagram of the application.

[0042] Figure 6 is the system code file framework schematic diagram of the application; in the figure: 1, weather.py obtains weather forecast parameters; 2, datproc.py and datmetc.py clean and analyze historical observation data; 3, datpred.py analyzes and screens suitable models, model training and prediction; 4, datplot.py and datfigs.py analyze prediction results and visualize data; 5, common.py is general data processing; 6, the corresponding *.sh file is a script program using xargs for multi-process parallel acceleration calculation. DETAILED DESCRIPTION

[0043] The application will be described in detail below in combination with theoretical derivation and formulas:

[0044] 1. Chemical reaction equilibrium principle

[0045] The total chemical equation of ozone reaction generation can be expressed as:

[0046]

[0047] Ozone concentration generation rate

[0048]

[0049] Generally speaking, organic matter tends to chemical reaction under certain conditions after a certain distance of transmission and reaches a quasi-equilibrium state. Therefore, it can be assumed that each substance in the atmosphere is close to chemical equilibrium Chemical equilibrium constant K = k p / k r , after rearrangement,

[0050]

[0051] Take the logarithm of both sides, and assume After rearrangement,

[0052]

[0053] In the formula: α i , δ k, ∈ are fitting coefficient terms, in the embodiments of the present application, the values are the results of training according to historical data of each site, there are many influencing factors and the data difference can be large; the coefficient α i (no dimension) can be positive or negative, generally less than 4 orders of magnitude (10 4 ); the coefficient δ k (no dimension) can be negative, generally less than 4 orders of magnitude (10 4 ); ∈ (no dimension) is a constant coupled with meteorological factors, generally less than 4 orders of magnitude (10 4 ). Here, the relationship between VOCs and ozone is applicable to various types of organic matter, including BVOC, OVOC, and organic halides.

[0054] 1.1 Relationship between ozone concentration and VOCs concentration

[0055] Therefore, the ozone concentration and the VOCs concentration are not in a linear relationship, but in a power function relationship, and the linear correlation relationship can be obtained after taking the double logarithm. That is,

[0056]

[0057] 1.2 Relationship between ozone concentration and NOx concentration

[0058] Similarly, the ozone concentration and the NOx concentration are not in a linear relationship, but in a power function relationship, and the linear correlation relationship can be obtained after taking the double logarithm. That is,

[0059]

[0060] 1.3 Relationship between ozone concentration and particulate matter concentration

[0061] It is assumed that the impurities such as heavy metals in particulate matter play a role of catalyst in the ozone generation process, and the increase of the particulate matter concentration means the increase of the specific surface area of the catalyst, and the specific surface area has a linear relationship with the reaction rate constant k p ,

[0062] k p = k p0 A = k' p0 C PM (7)

[0063] In the formula, A is the specific surface area of the catalyst, which is proportional to the particulate matter concentration.

[0064] Therefore, the chemical equilibrium constant K = k p / k r has a linear relationship with the particulate matter concentration, and the logarithmic value of the ozone concentration has a linear relationship with the logarithmic value of the particulate matter concentration. That is,

[0065]

[0066] wherein: a', β' are the fitting coefficients, in the embodiment of the present application, the values are trained according to the historical data of each station, and there are many influencing factors, so the data difference can be large. According to the influence characteristics of particulate matters in the region, the coefficient a' (dimensionless) can be positive or negative, mainly negative in the northern region and mainly positive in the southern region, generally less than 4 orders of magnitude (10 4 ), and β' (dimensionless) is generally less than 1 order of magnitude.

[0067] In addition, if it is assumed that particulate matters are generated by atmospheric chemical reaction process, the above relationship can also be derived by formula (4).

[0068] 1.4 Relationship between ozone concentration and air pressure

[0069] Since the gas state equation PV = nRT can be obtained,

[0070] C = PM / RT (9)

[0071] The concentration of each gas component has

[0072]

[0073] By substituting the reaction equilibrium constant calculation formula, we have

[0074]

[0075] Obviously, the logarithm of ozone and the logarithm of air pressure have a linear relationship, but this linear relationship is based on the premise that the sum of the chemical reaction equilibrium coefficients is not equal to zero, so the ozone concentration and the air pressure can also be irrelevant. That is,

[0076]

[0077] 2. Thermodynamic equilibrium principle

[0078] 2.1 Relationship between ozone concentration and humidity

[0079] Under standard atmospheric pressure, 1 volume of water dissolves 0.494 volume of ozone, and even ozone and water molecules can undergo the following irreversible reaction.

[0080] O3 + H2O → H2O2 + O2 (13)

[0081] The increase of humidity in the air will increase the probability of interaction between water molecules and ozone, so the ozone concentration will decrease. Both the concentration and the humidity are functions of the density of matter, and both have a linear relationship with the probability. Therefore, the ozone concentration in the air Linearly related to humidity W.

[0082]

[0083] In the formula, kW is the amount of dissolved ozone in the air, according to Henry's law, the ozone gas partial pressure p i = Hx i The molar fraction solubility of ozone dissolved in water x

[0084] p i = Hx i (15)

[0085] In the formula, the gas partial pressure p i = y i P. So when the pressure rises, the molar fraction solubility of ozone x i Must increase the amount of ozone dissolved in air water vapor, according to the principle of conservation of matter, the molar fraction of ozone in the air y i Must be reduced, the same humidity W conditions, the amount of dissolved ozone will increase, then k∝(P-P * ), therefore the concentration of ozone in the air With humidity W and air pressure P has the following relationship,

[0086]

[0087] In the formula: k', P * , b' are all fitting coefficient terms, in the embodiment of the application, the values are the results trained according to the historical data of each site, there are many influencing factors and the data difference may be large, k' is generally between ± 0.5 μg / m 3 , P * Is between-200 and-1500 hPa, b' has a large difference in the result of different combination formulas coupled with other variables, generally less than 3 orders of magnitude (10 3 ) μg / m 3 .

[0088] 2.2 Relationship between ozone concentration and temperature

[0089] The Gibbs-Helmholtz equation is derived from the basic equation of thermodynamics:

[0090]

[0091] Since Δ T G / T = -RlnK, there is

[0092]

[0093] The integral gives the relationship between the chemical equilibrium constant and the reaction enthalpy

[0094]

[0095] Therefore, the logarithm value of the ozone concentration has a linear relationship with the negative 1st power of the temperature. By substituting H = U + PV, the relationship between the ozone and the temperature T and the air pressure P can be further obtained as follows:

[0096]

[0097] In the formula, a, b, c, and d all refer to fitting coefficient terms, and in the embodiments of the present application, the values are results trained according to historical data of each station, and the influencing factors are many, and the data difference can be large, a is generally between -1500 and 500 ℃ / hPa, b can be a positive number or a negative number, and is generally between 0 and 6.5 orders of magnitude (unit: ℃), c (unitless) is generally between -2000 and 5000, and d (unitless) can be a positive number or a negative number, and is generally between -1 and 5 orders of magnitude; or a simplified formula in which any one or any two of the three terms of lnP is removed.

[0098] 3. Photochemical reaction principle

[0099] 3.1 Relationship between ozone concentration and ultraviolet radiation intensity

[0100] According to the Beer-Lambert law, the following relationship is obtained:

[0101] lgI o / I = εCl = A (21)

[0102] In the formula, I0 is the incident light intensity, I is the transmitted light intensity, C is the concentration of the substance absorbing light (mol / L), l is the solution thickness (cm), ε is the molar absorption coefficient, which reflects the characteristics of the light-absorbing substance and the possibility of electron transition, and A is the absorbance or optical density.

[0103] Photochemical efficiency Generally, the quantum yield is used to measure the photochemical efficiency, and the following relationship is obtained:

[0104]

[0105] The size of the quantum yield is related to the structure of the reactants and the reaction conditions (temperature, pressure, and concentration). The quantum yield of many photochemical reactions is between 0 and 1. In a chain reaction, one photon can trigger a series of chain reactions, and the quantum yield can reach several powers of 10. For example, the quantum yield of the free radical halogenation reaction of alkanes is 10 5 Under specific temperature and pressure conditions, this specific condition for the ozone generation reaction can be regarded as a constant. Through integration, the following formula is obtained:

[0106] ​

[0107] In the formula: I0 refers to the intensity of incident light; refers to the photochemical efficiency; C refers to the concentration of the substance absorbing light (mol / L); t refers to the photochemical reaction time, which in the embodiments of the present application is a result trained according to historical data of each site, and there are many influencing factors and the data difference can be large. I0 is generally within 3 orders of magnitude W / m 2 , Generally between -160 and 10 μg / m 3 , C is generally small, within 1000 μg / m 3 .

[0108] Therefore, the ozone concentration is proportional to the logarithm of the light intensity.

[0109] 4. Specific implementation examples

[0110] 4.1 Procedure implementation steps

[0111] (1) Obtain weather forecast parameters

[0112] ① Obtain weather forecast data from the website of China Meteorological Administration or the Central Meteorological Observatory, such as obtaining the weather forecast data of Fuzhou from the following two websites:

[0113] https: / / weather.cma.cn / web / weather / 58847.htm

[0114] http: / / www.nmc.cn / publish / forecast / AFJ / fuzhou.html

[0115] ② Parse the html data and read the precipitation, temperature, wind speed, wind direction, air pressure, humidity and cloud cover data, and store them for later use;

[0116] (2) Historical observation data cleaning and analysis

[0117] ① Read the obtained file data;

[0118] ② Clean invalid data and perform data statistical processing;

[0119] (3) Analysis and screening of suitable models

[0120] ① Select a model and execute "model training and prediction, prediction result analysis";

[0121] ② Select and determine the suitable model according to the prediction result;

[0122] (4) Data visualization

[0123] Visualize the prediction results, such as through graphical analysis.

[0124] 4.2 Verification of the Mechanism Model of Fujian Superstation Taking Fujian superstation as an example, historical data fitting analysis was performed on the key formulas (16) and (23). The data analysis results are shown in Figures 1-2 :

[0125] Figure 1 The analysis using historical data from Fujian Province's superstations yielded data that conformed to formula (16). ozone concentration The relationship between humidity (W) and air pressure (P) expresses the same pattern, and the calculations of data from other stations also follow the same pattern. Figure 1 The top left figure shows the locations colored using wind speed data, indicating that high ozone concentrations are mainly observed when wind speeds are between 1.0 and 3.3 m / s. Figure 1 The upper right figure shows the points colored using wind direction data, illustrating that high ozone concentrations are mainly observed in winds from southeast to southwest. Figure 1 The lower left figure in the image shows the points colored using temperature data; Figure 1 The lower right figure shows the coloring using air pressure data, illustrating that as air pressure increases, ozone levels decrease more rapidly with increasing humidity, demonstrating a significant coupling between humidity and pressure. In summary, from... Figure 1 As can be seen, when ozone concentration is high, the wind direction is concentrated in a few directions (southeast to southwest) between wind speeds of 1.0-3.3 m / s. When the temperature is greater than 17℃, the humidity is less than 70%, and mostly less than 60%, the air pressure is between 100-102 kPa.

[0126] Figure 2 This graph illustrates the effects of temperature and pressure coupling on ozone concentration, used to analyze the influence of air temperature (T) and air pressure (P) on ozone concentration. Figure 2 In the above figure, the shapes of the four points 1, 2, 3, and 4 represent data from spring, summer, autumn, and winter, respectively. They are marked with color using air pressure data. The two upward sloping lines are the ideal fitting line (the upper line, which is the data obtained in the summer and autumn seasons, i.e., the high temperature and low pressure zone) and the actual fitting line (the lower line). The data in the interval enclosed below the actual fitting line will have a higher ozone concentration in the winter and spring seasons, i.e., the low temperature and high pressure zone, and should be shifted to the left. Figure 2 In the figure below, the fitted curve R is obtained by removing the data from low temperature and high pressure (concentrated in winter and spring). 2 The value increased significantly, indicating that the ozone concentration was generally high under low temperature and high pressure (winter and spring). Therefore, the corrected formula (20) needs to be used. To predict and analyze.

[0127] 4.3 Overview of Fujian Province Superstation Forecasting

[0128] Take Fujian Super Station as an example for prediction and data analysis. First, using the historical data of Fujian Super Station from 2017 to 2018, the formulas (1)-(23) described in steps 1 to 3 are used for training, and then the prediction is made using the weather forecast data (temperature, pressure, humidity, etc.) obtained from the China Meteorological Administration website or the Central Meteorological Observatory in 2019, the historical data of ultraviolet radiation obtained from Fujian Super Station and Fuzhou Super Station, and the historical data of VOCs and NOx of the previous day, etc. Then the graph is obtained as follows Figures 3-4

[0129] Figure 3 is the comparison graph of the prediction results of the ozone concentration of the present application using the prediction models derived from formulas (4), (8), (16), (20), (23) and the actual values, which is used to evaluate the prediction error of the overall model, and a relatively accurate ozone concentration prediction model can be obtained by randomly selecting any combination of characteristic relationships, especially meteorological parameter combinations. The relationship between meteorological parameters, particulate matter, VOCs, NOx inorganic matter and ozone concentration can also be revealed from the mechanism. Figure 3 In the left graph in the above, u represents wind speed, and the prediction is more accurate when the wind speed is between 1-3 m / s, with an error of basically within 25%; in the right graph above, a represents wind direction, and the prediction is more accurate when the wind direction is between southeast and southwest, with an error of basically within 25%; Figure 3 In the left graph below, T represents temperature, and the prediction is more accurate when the temperature is between 15-30℃, with an error of basically within 25%; in the right graph below, P represents pressure, and the prediction is more accurate when the pressure is greater than 100 KPa, with an error of basically within 25%; Figure 3 In the left and right graphs above and below, the middle solid line is the line with an error of 0, and the upper and lower dashed lines represent the lines with an error of +25% and -25% respectively. From Figure 3 The comparison graph of the prediction results of the ozone concentration using the prediction models derived from claims 2-8 and the actual values can be obtained, and it can be concluded that the overall error of the model basically meets the requirements of HJ 1130-2020 "Technical Specifications for Numerical Prediction of Ambient Air Quality".

[0130] Figure 4 is the time series graph of the prediction results of the ozone concentration of the present application using the prediction models derived from formulas (4), (8), (16), (20), (23) and the actual values, which is used to measure the time series coincidence of the prediction results, and a relatively accurate ozone concentration prediction model can be obtained by randomly selecting any combination of characteristic relationships, especially meteorological parameter combinations. The relationship between meteorological parameters, particulate matter, VOCs, NOx inorganic matter and ozone concentration can also be revealed from the mechanism. Figure 4 The dashed line in is the prediction result, and the solid line is the actual value. From Figure 4 ​The comparison between the prediction results of the ozone concentration using the model derived in claims 2-8 and the actual values shows that the time series graph of the prediction results basically matches the time series graph of the actual situation.

[0131] 4.3 Overall forecast of coastal cities in the province

[0132] Take the coastal areas of Fujian Province as an example. A total of 25 station data were collected this time. Due to insufficient data, the available training data and prediction data are limited. The data of 22 stations are basically available, and the data of only 14 stations exceed 2 years. Among them, the two stations of Jinjiang No. 1 Middle School and Nan'an No. 1 Middle School have a large amount of missing meteorological data, and all indicators are not ideal. Only 20 stations can be implemented. Among them, only the super station of Fujian Province and the monitoring station of Xiamen City have sufficient data, so the analysis conclusions of these two stations are mainly used. The super station of Fujian Province and the monitoring station of Xiamen City use data from 2017 to 2018 for training, and other stations use data from 2018 to 2019. The super station of Fujian Province and the monitoring station of Xiamen City use data from 2019 for prediction, and other stations use data from 2020 for prediction. The meteorological parameters selected in the study include temperature, pressure, humidity, wind speed, and wind direction. Some stations further use particulate matter concentration for correction. The prediction results of this time use 5-level warning levels of 160 μg / m 3 , 150 μg / m 3 , 140 μg / m 3 , 130 μg / m 3 , and 120 μg / m 3 . See Tables 1-5 for specific prediction results.

[0133] As can be seen from Tables 1-5, among the 20 stations, the correlation coefficient is more than 0.75, the absolute value of the standardized average deviation is less than 0.25, and the root mean square error is less than 31. Among them, the super station of Fujian Province and the monitoring station of Xiamen City have relatively sufficient data and good consistency. The accuracy rate of light pollution weather reaches 75%, and the light pollution weather forecast test score (actual) reaches 75%. Among the 5 different warning levels of all stations, the accuracy rate of light pollution weather of nearly 1 / 2 stations and the light pollution weather forecast test score (actual) of nearly 1 / 3 stations can reach more than 70%. The algorithm involved in the model in this study is relatively less affected by the data volume. Some stations such as Quanzhou Fengze Qingyuan Mountain and Quanzhou Jinjiang Tushan Street still have relatively good results under the condition of lacking data.

[0134] Table 1 Prediction results of warning level 160 μg / m 3

[0135]

[0136] Table 2 Prediction results of warning level 150 μg / m​3 Predicted results

[0137]

[0138]

[0139] Table 3 Predicted results for a warning level of 140 μg / m 3 Predicted results

[0140]

[0141]

[0142] Table 4 Predicted results for a warning level of 130 μg / m 3 Predicted results

[0143]

[0144]

[0145] Table 5 Predicted results for a warning level of 120 μg / m 3 Predicted results

[0146]

Claims

1. An atmospheric environment ozone concentration prediction system based on chemical reaction mechanism, characterized by Comprise: 1) system software functions, including obtaining weather forecast parameters, historical observation data cleaning analysis, analysis and screening suitable models, model training and prediction, prediction result analysis, data visualization; the analysis and screening suitable model is specifically ① selecting a model, executing "model training and prediction, prediction result analysis"; ② selecting and screening to determine suitable model according to the prediction result; 2) each software function module of the system is independently designed, which can be independently run or parallel accelerated calculation, and can be single machine processing or easily deployed cluster calculation and cloud calculation; 3) based on the Lagrange coordinate system, the response relationship between meteorological parameters, pollutant concentration and ozone concentration is established; 4) using the theory of thermodynamic equilibrium principle, chemical reaction equilibrium principle and photochemical reaction principle basic chemical reaction engineering, a nonlinear model of the influence of meteorological parameters and pollutant concentration parameters on ozone concentration is derived.

2. The atmospheric environment ozone concentration prediction system based on chemical reaction mechanism according to claim 1, characterized in that The prediction model of the ozone concentration can characterize the humidity and the air pressure coupling the influence of the ozone concentration satisfying the relationship wherein: takes values between ± 0.5 μg / m 3 ; takes values between -200 ~ -1500 hPa; denotes the fitting coefficient term.

3. The atmospheric environment ozone concentration prediction system based on chemical reaction mechanism according to claim 1, characterized in that The prediction model of the ozone concentration can characterize the influence of the air temperature and the air pressure coupling on the ozone concentration , and satisfy a * ln (P) + b = c * ln (T) + d, in which: a has a value between -1500 and 500 °C / hPa; b has a value between 0 and 6.5 orders of magnitude, in °C; c has a value between -2000 and 5000; d has a value between -1 and 5 orders of magnitude.

4. The atmospheric environment ozone concentration prediction system based on chemical reaction mechanism according to claim 1, characterized in that The prediction model of the ozone concentration can characterize the intensity of solar radiation the influence of the ozone concentration , meeting A is the absorbance or optical density, I is the incident light intensity; E is the photochemical efficiency; C is the concentration of the light-absorbing substance in mol / L; and t is the photochemical reaction time.

5. The atmospheric environment ozone concentration prediction system based on chemical reaction mechanism according to claim 1, characterized in that The prediction model of the ozone concentration can represent the relationship between the concentration of volatile organic compounds VOCs , nitrogen oxides inorganic compounds and the ozone concentration , satisfying where: values less than 4 orders of magnitude (10 4 ); values less than 4 orders of magnitude (10 4 ); values less than 4 orders of magnitude (10 4 ); here VOCs' relationship with ozone applies to all types of organics, including BVOCs, OVOCs, and organic halides.

6. The atmospheric environment ozone concentration prediction system based on chemical reaction mechanism according to claim 1, characterized in that The prediction model of the ozone concentration can represent the relationship of the particulate matter concentration with the ozone concentration , and satisfy in relation to the formula where: values less than 4 orders of magnitude (10 4 ); values less than 1 order of magnitude.

Images