Prediction Of PM_2.5 Concentration In Tianjin Based On Spatial Correlation

In recent years,while people are more and more pursuing the quality of life,the problem of air pollution is becoming more and more serious.Severe haze weather in many parts of our country has put people’s health at great risk,especially the impact on the human respiratory system,which has seriously lowered the The happiness and sense of achievement of the general public.The fog in the haze is not harmful to the human body.The haze refers to the suspended particles in the air.Among them PM_2.5 due to its own characteristics,compared with other particulate matter,the damage to the human body is greater,and PM_2.5 has become one of the most important factors that have a serious impact on people’s production,life and even survival.Finding a way to predict the PM_2.5 concentration value more accurately is very necessary.In this paper,aiming at the prediction of PM_2.5 concentration in Tianjin,based on the air pollution and meteorological data in the Beijing-Tianjin-Hebei region in the past five years（January 1,2016 to June 30,2021）,some new combined model methods are established.And successfully predict PM_2.5 concentration.Firstly,the ARMA model was constructed by using the historical data of PM_2.5,and the multivariate time series model ARIMAX was constructed by adding air pollution and meteorological data,and the optimal feature selection method was compared and selected.The recursive feature elimination algorithm（RFE）reduces the variable dimension and improves the prediction accuracy.and operational efficiency.Secondly,based on the establishment of a single time series model,in order to solve the problem that the noise in the air quality and meteorological observation data can easily mask the actual trend of PM_2.5,this paper firstly combines the ensemble empirical mode decomposition（EEMD）and The combined model of the time series model ARIMAX was used in PM_2.5 concentration prediction.Finally,considering that each area is not closed,and the change of air quality in the neighborhood is also a factor that cannot be ignored in affecting the PM_2.5 concentration in this area,the spatial correlation is added on the basis of the above model,and the impact of adjacent areas on Tianjin is considered,synthesizing the horizontal and vertical influences,and establishing a spatial correlation-based EEMD-ARIMAX prediction model.The results of this paper show that when the model properly applies the influence of external factors and the information in the data is more regular,it shows a better prediction effect.The multivariate combination model can not only make full use of the information contained in the exogenous variables,but also fully ensure the sequence stationarity through EEMD,and reduce the influence of noise on the model establishment;comprehensively considering the influence between regions can effectively improve the prediction accuracy.In terms of policy,from the perspective of the spatial dependence of my country’s haze pollution and the spatial distribution characteristics of regional agglomerations,solving the problem of air pollution requires coordination and cooperation between provinces,cities and regions.In terms of production and life,the forecast results of PM_2.5 combined with the new air quality standards can provide reference suggestions for people’s travel and activities to protect people’s health.