O3 pollution prediction method and system based on multiple linear regression

Abstract

The present invention relates to the technical field of ambient air quality forecasting, specifically to an O3 pollution forecasting method and system based on multiple linear regression. The O3 pollution forecasting method based on multiple linear regression of the present invention comprises the following steps: S1, acquiring data; S2, preprocessing the data and constructing an O3 pollution case library; S3, classifying O3 pollution cases according to meteorological elements; S4, screening appropriate meteorological elements as forecasting factors for each type of O3 pollution case; S5, establishing a stepwise multiple linear regression model of meteorological data and pollutant concentration for each type of O3 pollution case, and evaluating the regression effect; S6, obtaining the forecast results of future daily weather elements using a high-resolution atmospheric chemistry numerical model, and using the multiple linear regression model to predict future O3 pollution conditions. The O3 pollution event forecasting method provided by the present invention has a simple calculation method and a high accuracy of the output forecast results, effectively improving the effectiveness of O3 pollution forecasting and providing an open research framework for O3 pollution forecasting, providing strong support for related research such as the prediction and evaluation of O3 complex pollution.

Description

Technical Field

[0001] The present invention relates to the technical field of ambient air quality forecasting, and in particular to an O3 pollution forecasting method and system based on multiple linear regression. Background Art

[0002] With the accelerated pace of industrialization and urbanization, air pollution is becoming increasingly severe, becoming a major obstacle to sustainable development. Ozone (O3), a major atmospheric pollutant, poses a threat to human health and damages ecosystems and crops at elevated concentrations. Therefore, accurately predicting ozone concentrations and studying its regional transport mechanisms are crucial for developing effective air pollution prevention and control measures and improving air quality.

[0003] Traditional ozone pollution forecasting methods primarily rely on chemical transport models, such as WRF-CMAQ. While these methods can simulate and predict ozone pollution to a certain extent, they suffer from limitations such as high computational complexity, poor timeliness, and incomplete representation of physical and chemical processes. The accuracy and applicability of these traditional models are particularly limited when dealing with nonlinear problems.

[0004] In recent years, the rapid development of big data and machine learning technologies has provided new insights and methods for ozone pollution forecasting. Machine learning models, with their adaptability, high accuracy, and low computational complexity, have shown great potential in addressing nonlinear problems and forecasting complex systems. By leveraging extensive historical meteorological and air quality data, machine learning models can more accurately capture the nonlinear characteristics of ozone concentration fluctuations, improving both the accuracy and timeliness of forecasts.

[0005] Multiple linear regression, a classic statistical analysis method, has considerable application value in ozone pollution forecasting. It analyzes the linear relationship between ozone concentration and meteorological parameters to establish a prediction model, enabling prediction of future ozone concentrations. However, multiple linear regression has limitations in handling nonlinear problems and complex system predictions, and its prediction accuracy and generalization capabilities need to be improved. Summary of the Invention

[0006] The purpose of the present invention is to provide an O3 pollution prediction method and system based on multiple linear regression.

[0007] In a first aspect, the present invention provides an O3 pollution prediction method based on multiple linear regression, which specifically comprises the following steps:

[0008] S1. Obtain information;

[0009] S2. Preprocess the data and build an O3 pollution case database;

[0010] S3. Classify O3 pollution cases based on meteorological factors;

[0011] S4. Select appropriate meteorological factors as predictors for each type of O3 pollution case;

[0012] S5. For each type of O3 pollution case, establish a stepwise multiple linear regression model between meteorological data and pollutant concentrations and evaluate the regression effect;

[0013] S6. Obtain forecasts of future daily weather elements using high-resolution atmospheric chemistry numerical models, and use multivariate linear regression models to predict future O3 pollution.

[0014] In one embodiment of the present application, in S1, the acquired data include pollutant concentration site data and reanalysis data grid data; wherein the pollutant concentration site data include: O3 concentration site data, with a time resolution of 1h; and the reanalysis data grid data include: boundary layer height, 2m dew point temperature, 2m temperature, precipitation, sea level pressure, surface shortwave radiation, cloud cover, geopotential height, relative humidity, specific humidity, radial wind field, latitudinal wind field, vertical wind field, with a spatial resolution of 0.25°×0.25° and a time resolution of 1h.

[0015] In one embodiment of the present application, in S2, the step of preprocessing the data and building an O3 pollution case library includes:

[0016] S2.1. Calculate the maximum 8-hour moving average of the daily O3 concentration:

[0017]

[0018] in, is the 8-hour average of the O3 concentration at a certain station starting from the i-th hour on the same day; O3(i) is the O3 concentration at a certain station at the i-th hour on the same day; n1 is the amount of valid data. If the valid data within the 8-hour period is less than 6 hours, it will not be counted in the case; MDA8O3 is the maximum 8-hour sliding mean of the O3 concentration at a certain station, and n2 is the amount of valid data.

[0019] S2.2. Calculate the daily maximum 2-meter temperature, daytime boundary layer height, daytime average temperature, daytime surface shortwave radiation, and daytime cloud cover;

[0020] S2.3. Screening for the maximum 8-hour sliding mean of O3 concentration exceeding 160 μg m -3 The O3 pollution cases constitute the pollution case database.

[0021] In one embodiment of the present application, in S3, the self-organizing map method is used for typing, and the steps include:

[0022] S3.1. For a given region, standardize the 850 hPa height field grid data:

[0023]

[0024] Where Z is the original data of the 850hPa height field; μ is the mean value of the training data; σ is the standard deviation of the training data; It is the standardized data of 850hPa height field;

[0025] S3.2. Establish and initialize the SOM network, including: setting the number of nodes m in the output layer based on the number of weather types required, establishing an initial winning neighborhood and an initial value for the learning rate α, and randomly assigning a weight vector to each node in the output layer, i.e., each grid node j has a weight vector with the same dimension as the input feature vector;

[0026] S3.3. Finding the winning node includes: for a randomly sampled input sample Z, finding the closest node by calculating the weight vector difference between the input sample and all nodes. This node is the winning node, namely the "Best Matching Unit" BMU, whose index is b. The BMU must meet the following requirements:

[0027] BMU=argmin j ||ZW j ||

[0028] Among them, Z is the current sample vector, that is, the standardized data of the 850hPa height field, and W i is the weight vector of each node, ||·|| is the Euclidean distance EDU, and EDU must satisfy:

[0029]

[0030] Where n is the number of neurons in the input layer, i.e., the sample size of the 850 hPa height field;

[0031] S3.4. Update the weight vectors of the BMU and each node in its domain to make them closer to sample Z. The weight update must satisfy:

[0032]

[0033] Among them, t is the current iteration variable, W j (t) is the weight of node j at the tth iteration, α(t) is the learning rate at the tth iteration, h(t) is the neighborhood function value, and the Gaussian neighborhood function is selected, which must satisfy:

[0034]

[0035] Among them, h is the neighborhood function value, which represents the update weight of the nodes near the BMU, ║rb -r j ║ is the grid coordinate Euclidean distance between the BMU and the surrounding nodes, and σ is the neighborhood width.

[0036] S3.5. Repeat the above process several times until the training is completed.

[0037] In one embodiment of the present application, in S4, the step of selecting appropriate meteorological elements as forecasting factors for each type of O3 pollution case includes:

[0038] S4.1. Calculate the Pearson correlation coefficient R between the MDA8 O3 concentration of each pollution case and the candidate meteorological variables in the reanalysis data grid data. Meteorological variables that are statistically significant at the 95% confidence level are retained as candidate meteorological characteristic factors for the next selection step. They must meet the following requirements:

[0039]

[0040] in, and is the average value of the related time series x and y;

[0041] S4.2. Use the gradient boosting decision tree algorithm to further screen meteorological characteristic factors based on the importance of eigenvalues.

[0042] In one embodiment of the present application, in S5, a stepwise multiple linear regression model of meteorological data and pollutant concentrations is established for each type of composite pollution case, and the step of evaluating the regression effect includes:

[0043] S5.1 Constructing a stepwise multiple linear regression model MLR requires the following:

[0044]

[0045] in, For MLR prediction of pollutant concentration, b0 represents the intercept term, b k represents the regression coefficient, M k represents the predictor, the total number is N;

[0046] Add or delete predictors based on the Akaike information classification statistic and perform stepwise regression, which requires:

[0047]

[0048] Among them, SSE is the difference between the pollutant concentration C and the MLR predicted pollution concentration The sum of squared errors between ; T is the number of days, and N is the number of predictors used in the regression;

[0049] The relative weight of each predictor on pollutant concentration is measured by its relative contribution to the total explained variance of the multiple linear regression, which must satisfy:

[0050]

[0051] Among them, w k is the weight of the predictor, N is the number of predictors, b k represents the regression coefficient, S k represents the standard deviation of the prediction factor i, S0 represents the standard deviation of the pollutant concentration;

[0052] S5.2. Evaluate the regression effect; perform a significance test on the overall regression equation to determine whether all independent variables have a significant linear relationship with the dependent variable. This is done using the F test. The F statistic is the ratio of the regression sum of squares to the error sum of squares and must meet a 95% significance level:

[0053]

[0054] In one embodiment of the present application, in S6, the method of obtaining the forecast results of future daily weather elements of the atmospheric chemistry numerical model with high resolution, and predicting the future O3 pollution situation using the multivariate linear regression model includes: obtaining the daily meteorological data of the required prediction factors, and inputting them into the determined multivariate linear regression equation to calculate the future pollutant concentration of the city.

[0055] In one embodiment of the present application, the O3 pollution prediction method further includes: S7, reviewing and summarizing the O3 pollution process and improving the case library.

[0056] In one embodiment of the present application, in step S7, the review and summary content includes: actual pollution situation, pollution forecast and forecast verification results.

[0057] In a second aspect, the present invention provides an O3 pollution prediction system based on multiple linear regression, which adopts the O3 pollution prediction method based on multiple linear regression as described above, comprising:

[0058] A data acquisition module, used to acquire data;

[0059] The pollution case library construction module is used to pre-process the data and build an O3 pollution case library;

[0060] Classification module, used to classify O3 pollution cases based on meteorological factors;

[0061] Prediction factor screening module, used to select appropriate meteorological elements as prediction factors for each type of O3 pollution case;

[0062] The model building module is used to establish a stepwise multiple linear regression model between meteorological data and pollutant concentrations for each type of O3 pollution case and evaluate the regression effect;

[0063] The O3 pollution forecast module is used to obtain the forecast results of future daily weather elements using high-resolution atmospheric chemistry numerical models and predict future O3 pollution using a multivariate linear regression model;

[0064] The review and improvement module is used to review and summarize the O3 pollution process and improve the case library.

[0065] Compared with the prior art, the present invention has the following beneficial effects:

[0066] The O3 pollution forecasting method based on multiple linear regression of the present invention adopts long-term station data and reanalysis data with high temporal and spatial resolution, combines the nonlinear processing capability of machine learning and the stability of multiple linear regression, and can more accurately capture the complex relationship between ozone concentration and various influencing factors, and can effectively supplement the ozone generation, elimination and transport characteristics, thereby improving the accuracy and timeliness of the prediction; at the same time, the method can also comprehensively consider more types of data, providing strong support for the accurate forecast and prevention and control of ozone pollution. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0068] Figure 1 Schematic diagram of the O3 pollution prediction method based on multiple linear regression of the present invention;

[0069] Figure 2 The actual results of MDA8 O3 and pollution days in the Yangtze River Delta region from 2015 to 2023 according to one embodiment of the present invention are as follows;

[0070] Figure 3 This is the weather classification result for the Yangtze River Delta region in 2019 according to one embodiment of the present invention. The coloring indicates the 850hPa height field, the arrows indicate the wind field, and the colors represent the temperature.

[0071] Figure 4 This is a line graph showing the change of MDA8 O3 over time in the Yangtze River Delta region from July 24, 2020 to July 31, 2020, according to an embodiment of the present invention. The solid line represents the observed value, and the dotted line represents the forecast value. DETAILED DESCRIPTION

[0072] The technical solution of the present invention is described in detail below with reference to the accompanying drawings and specific implementation cases. It should be noted that the embodiments described here are only for illustration and are not intended to limit the present invention.

[0073] In one embodiment, if Figure 1 As shown in FIG, the O3 pollution prediction method based on multiple linear regression includes the following steps:

[0074] S1. Obtain information;

[0075] S2. Preprocess the data and build an O3 pollution case database;

[0076] S3. Classify O3 pollution cases based on meteorological factors;

[0077] S4. Select appropriate meteorological factors as predictors for each type of O3 pollution case;

[0078] S5. For each type of O3 pollution case, establish a stepwise multiple linear regression model between meteorological data and pollutant concentrations and evaluate the regression effect;

[0079] S6. Obtain forecasts of future daily weather elements using high-resolution atmospheric chemistry numerical models, and use multivariate linear regression models to predict future O3 pollution.

[0080] Specifically, in S1, the data obtained include pollutant concentration site data and reanalysis data grid data. Pollutant concentration site data include: O3 concentration site data, the sites can include 391 sites across the country, with a time resolution of 1 hour; reanalysis data grid data may include: boundary layer height, 2m dew point temperature, 2m temperature, precipitation, sea level pressure, surface shortwave radiation, cloud cover, geopotential height, relative humidity, specific humidity, radial wind field, zonal wind field, vertical wind field, spatial resolution 0.25°×0.25°, time resolution 1 hour. Optionally, the data can be the fifth generation global weather and climate reanalysis data ERA5 from the European Centre for Medium-Range Weather Forecasts (ECMWF).

[0081] Furthermore, in S2, the step of pre-processing the data and building an O3 pollution case database includes:

[0082] The maximum 8-hour moving average of O3 concentration (MDA8 O3) exceeds 160 μg m -3 .

[0083] S2.1. Calculate the maximum 8-hour moving average of the daily O3 concentration using the following formula:

[0084]

[0085] in, is the 8-hour average of the O3 concentration at a certain station starting from the i-th hour on the day, O3(i) is the O3 concentration at a certain station at the i-th hour on the day, n1 is the amount of valid data, when the valid data within this 8-hour period is less than 6 hours (n1<6), it is not counted in the case, MDA8 O3 is the maximum 8-hour sliding mean of the O3 concentration at a certain station, and n2 is the amount of valid data.

[0086] S2.2. Calculate the daily maximum 2-meter temperature, daytime boundary layer height (8:00-17:00), daytime average temperature (8:00-17:00), daytime surface shortwave radiation (8:00-17:00), and daytime cloud cover (8:00-17:00);

[0087] S2.3. Screening for the maximum 8-hour sliding mean of O3 concentration exceeding 160 μg m -3 The O3 pollution cases constitute the pollution case database.

[0088] It can output O3 pollution actual data set and forecast data set, including the location, duration and process of O3 pollution events. Figure 2 , which shows the actual results of MDA8 O3 and polluted days (OPdays) in the Yangtze River Delta region from 2015 to 2023 in one embodiment. MAM: spring; JJA: summer; SON: autumn; DJF: winter.

[0089] Furthermore, in S3, a self-organizing map (SOM) method is used to perform weather classification, and the steps include:

[0090] S3.1. For a given area, the 850 hPa height field grid data are standardized using the following formula:

[0091]

[0092] Where Z is the original data of the 850 hPa height field (the eigenvalue of the individual sample); μ is the mean value of the training data (each column of eigenvalues); σ is the standard deviation of the training data (each column of eigenvalues); It is the standardized data of 850hPa height field;

[0093] S3.2. Establish and initialize the SOM network, including: setting the number of nodes m in the output layer (i.e., determining the number of weather types as needed), establishing the initial winning neighborhood and the initial value of the learning rate α, and randomly assigning a weight vector to each node in the output layer, i.e., each grid node j has a weight vector with the same dimension as the input feature vector;

[0094] S3.3. Finding the winning node includes: for a randomly sampled input sample Z, finding the closest node by calculating the weight vector difference between the input sample and all nodes. This node is the winning node, namely the "Best Matching Unit" BMU, whose index is b. The BMU calculation formula is as follows:

[0095] BMU=argmin j ║ZW j ║

[0096] Among them, Z is the current sample vector, that is, the standardized data of the 850hPa height field, and W i is the weight vector of each node, ||·|| is the Euclidean distance EDU, and the EDU calculation formula is as follows:

[0097]

[0098] Where n is the number of neurons in the input layer, that is, the sample size of the 850hPa height field.

[0099] S3.4. Update the weight vectors of the BMU and each node in its domain to make them closer to sample Z. The weight update must satisfy:

[0100]

[0101] Among them, t is the current iteration variable, W j (t) is the weight of node j at the tth iteration, α(t) is the learning rate at the tth iteration, h(t) is the neighborhood function value, and the Gaussian neighborhood function is selected, which must satisfy:

[0102]

[0103] Among them, h is the neighborhood function value, which represents the update weight of the nodes near the BMU, ║r b -r j ║ is the grid coordinate Euclidean distance between the BMU and the surrounding nodes, and σ is the neighborhood width.

[0104] S3.5. Repeat the above process (random sampling, winning node search, weight update) several times until the training is completed. The general training termination condition is: the learning rate is less than 0 or the specified number of iterations is reached.

[0105] Figure 3 This is the weather classification result of the Yangtze River Delta region in 2019 in one embodiment. The color is the 850hPa height field, the arrows are the wind field, and the colors represent the temperature.

[0106] Furthermore, in S4, the step of selecting appropriate meteorological elements as forecasting factors for each type of O3 pollution case includes:

[0107] S4.1. Calculate the Pearson correlation coefficient R between the MDA8 O3 concentration of each pollution case and the candidate meteorological variables in the reanalysis data grid data. Meteorological variables that are statistically significant at the 95% confidence level are retained as candidate meteorological characteristic factors for the next selection step. The formula is as follows:

[0108]

[0109] in, and is the mean of the related time series x and y.

[0110] S4.2. Use the gradient boosting decision tree algorithm to further screen meteorological characteristic factors based on the importance of eigenvalues.

[0111] In S5, a stepwise multiple linear regression model of meteorological data and pollutant concentrations is established for each type of complex pollution case, and the steps of evaluating the regression effect include:

[0112] S5.1. Construct a stepwise multiple linear regression model MLR, the formula is as follows:

[0113]

[0114] in, For MLR prediction of pollutant concentration, b0 represents the intercept term, b k represents the regression coefficient, M k Represents predictors, with a total number of N.

[0115] Add or delete predictors based on the Akaike information classification statistic and perform stepwise regression. The formula is as follows:

[0116]

[0117] Among them, SSE is the difference between the pollutant concentration C and the MLR predicted pollution concentration The sum of squared errors between . T is the number of days and N is the number of predictors used in the regression.

[0118] The relative weight of each predictor on pollutant concentration is measured by its relative contribution to the total explained variance of the multiple linear regression, as follows:

[0119]

[0120] Among them, w k is the weight of the predictor, N is the number of predictors, bk represents the regression coefficient, S k represents the standard deviation of the prediction factor i, and S0 represents the standard deviation of the pollutant concentration.

[0121] S5.2. The significance test of the overall regression equation is used to determine whether all independent variables have a significant linear relationship with the dependent variable. This is usually done using the F test. The F statistic is the ratio of the regression sum of squares to the error sum of squares and must meet the 95% significance level. The formula is as follows:

[0122]

[0123] Furthermore, in S6, the forecast results of future daily weather elements using a high-resolution atmospheric chemistry numerical model are obtained, and the future O3 pollution situation is predicted using a multiple linear regression model. The daily meteorological data of the required prediction factors are obtained and input into the determined multiple linear regression equation to calculate the future pollutant concentrations in the city.

[0124] Figure 4 This is a line graph showing the change of MDA8 O3 over time in the Yangtze River Delta region from July 24, 2020 to July 31, 2020, in one embodiment. The solid line represents the observed value, and the dotted line represents the predicted value.

[0125] Furthermore, the O3 pollution prediction method further includes: S7, reviewing and summarizing the O3 pollution process and improving the case library. Specifically, the prediction results are combined with the actual situation to review and summarize the O3 pollution process and improve the case library. The review and summary content includes: the actual pollution situation (spatiotemporal distribution characteristics); the pollution forecast (spatiotemporal distribution characteristics); and the forecast verification results.

[0126] In summary, the present invention uses long-term station data and reanalysis data with high temporal and spatial resolution, combines the nonlinear processing capabilities of machine learning and the stability of multivariate linear regression, and can more accurately capture the complex relationship between ozone concentration and various influencing factors. It can effectively supplement the ozone generation, disappearance and transport characteristics, thereby improving the accuracy and timeliness of the prediction; at the same time, this method can also comprehensively consider more types of data, providing strong support for the accurate prediction and prevention and control of ozone pollution.

Claims

1. A method for predicting O3 pollution based on multiple linear regression, characterized in that: The following steps are involved: S1. Obtain information; S2. Preprocess the data and build an O3 pollution case database; S3. Classify O3 pollution cases based on meteorological factors; S4. Select appropriate meteorological factors as predictors for each type of O3 pollution case; S5. For each type of O3 pollution case, establish a stepwise multiple linear regression model between meteorological data and pollutant concentrations and evaluate the regression effect; S6. Obtain the forecast results of future daily weather elements using high-resolution atmospheric chemistry numerical models, and use multiple linear regression models to predict future O3 pollution; In S3, the self-organizing map method is used for typing, and the steps include: S3.

1. For a given region, standardize the 850 hPa height field grid data: in, It is the original data of 850hPa height field; is the mean of the training data; is the standard deviation of the training data; It is the standardized data of 850hPa height field; S3.

2. Establish and initialize the SOM network, including: setting the number of nodes in the output layer according to the number of weather types required , establish the initial winning neighborhood and learning rate Initial value, randomly assign a weight vector to each node of the output layer, that is, each grid node There is a weight vector with the same dimension as the input feature vector; S3.3, Find the winning node, including: for the random sampling input sample By calculating the weight vector difference between the input sample and all nodes, the node with the closest distance is found. This node is the winning node, namely the "Best Matching Unit" BMU, whose index is b. The BMU must meet the following requirements: in, is the current sample vector, i.e. the standardized data of the 850hPa height field, is the weight vector of each node, It is the Euclidean distance EDU, EDU must satisfy: Where n is the number of neurons in the input layer, i.e., the sample size of the 850 hPa height field; S3.4, update the weight vectors of each node in the BMU and its domain so that they are closer to the sample To get closer, the weight update must satisfy: in, is the current iteration variable, is a node exist The weight at the iteration, is The learning rate at the iteration, is the neighborhood function value. The Gaussian neighborhood function is selected and must satisfy: in, is the neighborhood function value, which represents the updated weight of the nodes near the BMU. is the grid coordinate Euclidean distance between BMU and surrounding nodes, is the neighborhood width; S3.

5. Repeat the above process several times until the training is completed.

2. The O3 pollution forecasting method according to claim 1, characterized in that: In said S1, the acquired data include pollutant concentration station data and reanalysis data grid data; in The pollutant concentration site data includes: O3 concentration site data, with a time resolution of 1 hour; and The gridded data of the reanalysis data include: boundary layer height, 2-meter dew point temperature, 2-meter temperature, precipitation, sea level pressure, surface shortwave radiation, cloud cover, geopotential height, relative humidity, specific humidity, radial wind field, zonal wind field, and vertical wind field, with a spatial resolution of 0.25°×0.25° and a temporal resolution of 1 hour.

3. The O3 pollution forecasting method according to claim 1, characterized in that: In S2, the steps of pre-processing the data and building an O3 pollution case database include: S2.

1. Calculate the maximum 8-hour moving average of the daily O3 concentration: in, The O3 concentration at a certain station on that day The 8-hour average value starting from the time; The O3 concentration at a certain station on that day Time value; The amount of valid data. If the valid data in the 8-hour period is less than 6 times, it will not be counted as a case. is the maximum 8-hour sliding mean of the daily O3 concentration at a certain station, is the effective data volume; S2.

2. Calculate the daily maximum 2-meter temperature, daytime boundary layer height, daytime average temperature, daytime surface shortwave radiation, and daytime cloud cover; S2.

3. Screening: The maximum 8-hour moving average of the daily O3 concentration exceeds 160 μg m − 3 The O3 pollution cases constitute the pollution case database.

4. The O3 pollution forecasting method according to claim 2, characterized in that: In S4, the step of selecting appropriate meteorological elements as forecasting factors for each type of O3 pollution case includes: S4.

1. Calculate the Pearson correlation coefficient (R) between the MDA8 O3 concentration for each pollution case and the candidate meteorological variables in the reanalysis grid data. Meteorological variables that are statistically significant at the 95% confidence level are retained as candidate meteorological characteristic factors for the next selection step. They must meet the following requirements: in, and For the relevant time series and The average value of S4.

2. Use the gradient boosting decision tree algorithm to further screen meteorological characteristic factors based on the importance of eigenvalues.

5. The O3 pollution forecasting method according to claim 1, characterized in that: In S5, a stepwise multiple linear regression model of meteorological data and pollutant concentrations is established for each type of complex pollution case, and the steps of evaluating the regression effect include: S5.1 Constructing a stepwise multiple linear regression model MLR requires the following: in, To predict pollutant concentrations for MLR, represents the intercept term, represents the regression coefficient, Denotes the predictor, with a total of ; Add or delete predictors based on the Akaike information classification statistic and perform stepwise regression, which requires: Where SSE is the pollutant concentration Predicting pollution concentration with MLR The sum of squared errors between is the number of days, N is the number of predictors used in the regression; The relative weight of each predictor on pollutant concentration is measured by its relative contribution to the total explained variance of the multiple linear regression, which must satisfy: in, is the weight of the predictor, is the number of predictors, represents the regression coefficient, Represents the predictor The standard deviation of represents the standard deviation of pollutant concentration; S5.

2. Evaluate the regression effect; perform a significance test on the overall regression equation to determine whether all independent variables have a significant linear relationship with the dependent variable. This is done using the F test. The F statistic is the ratio of the regression sum of squares to the error sum of squares and must meet a 95% significance level: 。 6. The O3 pollution forecasting method according to claim 1, characterized in that: In said S6, the method of obtaining the forecast results of future daily weather elements by a high-resolution atmospheric chemistry numerical model and predicting the future O3 pollution situation by using a multiple linear regression model includes: obtaining daily meteorological data of the required prediction factors, and inputting them into a determined multiple linear regression equation to calculate the future pollutant concentration of the city.

7. The O3 pollution forecasting method according to claim 1, characterized in that: Also includes: S7. Review and summarize the O3 pollution process and improve the case library.

8. An O3 pollution prediction system based on multiple linear regression using the O3 pollution prediction method based on multiple linear regression according to any one of claims 1 to 7, characterized in that: include: A data acquisition module, used to acquire data; The pollution case library construction module is used to pre-process the data and build an O3 pollution case library; Classification module, used to classify O3 pollution cases based on meteorological factors; Prediction factor screening module, used to select appropriate meteorological elements as prediction factors for each type of O3 pollution case; The model building module is used to establish a stepwise multiple linear regression model between meteorological data and pollutant concentrations for each type of O3 pollution case and evaluate the regression effect; The O3 pollution forecast module is used to obtain the forecast results of future daily weather elements using high-resolution atmospheric chemistry numerical models and predict future O3 pollution using a multivariate linear regression model; The review and improvement module is used to review and summarize the O3 pollution process and improve the case library.

Images