Research On An Air Quality Forecasting Method Based On Stacking Ensemble Strategy In Beijing,China

Nowadays air pollution has become one of the most serious environmental problems in the world.Fine particulate matters(PM _2.5)prod particular threats to ambient air quality,economic development and human health.Considering Beijing and six surrounding cities as main research areas,this study takes the daily average pollutant concentrations and meteorological elements from December 2^nd,2013 to October 13^th,2017 into account,and studies the spatial and temporal distribution characteristics,the primary influencing factors and the forecasting method of PM_2.5 concentrations in Beijing in order to provide guidance for coping with extreme meteorological disasters,and to provide references for improving municipal crisis response and emergency planning.In this paper,the interannual,seasonal and diurnal variation trends and temporal spatial distribution characteristics of PM _2.5 concentration in Beijing are studied by correlation analysis and geostatistics.Special feature extraction procedures,including those of simplification,polynomial,transformation and combination,are conducted before modeling to identify potentially significant features based on an exploratory data analysis.Stability feature s election and tree-based feature selection methods are applied to select important variables and evaluate the degrees of feature importance.Single models including LASSO,Adaboost,XGBoost and multi-layer perceptron optimized by the genetic algorithm（GA-MLP）are established in the level 0 space and are then integrated by support vector regression（SVR）in the level 1 space via stacked generalization.The models performance are evaluated and compared under different metrics.The main conclusions are as follows:（1）The pollutant concentrations in Beijing exhibit obvious seasonal and cyclical fluctuation patterns.Air pollution is more serious in winter and spring,and slightly better in summer and autumn with the spatial distribution of pollutants fluctuating dramatically in different seasons.The pollutions in southern Beijing areas are more grievous,and the air quality in northern areas are better in general.The diurnal variation of air quality shows a typically seasonal difference and the daily variation of PM_2.5 concentrations basically presented the"W"type of mode with twin peaks.Except for the emissions and accumulations of local pollutants,air quality is susceptible to the transport effect from southwest.（2）Our model evaluation shows that the proposed ensemble model generally performs approximately 20%better than a single nonlinear forecasting model when applied to new data.For single pollutant grade recognition,the proposed model performs better when applied to days characterized by good air quality than when applied to days registering high levels of pollution.（3）A feature importance analysis reveals that nitrogen dioxide（NO ₂）and carbon monoxide（CO）concentrations measured from the city of Zhangjiakou are taken as the most important elements of pollution factors for forecasting PM_2.5concentrations.The concentrations of PM₁₀ and PM_2.5 are the most significant local influencing factors to Beijing air quality.Extreme wind speeds and maximal wind speeds are considered to extend the most ef fects of meteorological factors to the cross-regional transportation of contaminants.Pollutants found in the cities of Zhangjiakou and Chengde have a stronger impact on air quality in Beijing than other surrounding factors.Each element affects the air qu ality of study areas in a different way.In this paper,the influences of pollutant factors,meteorological elements and transport effects are comprehensively taken into consideration,and a novel air quality forecasting method is proposed through data exploration,feature extraction,feature selection and model integration.The results demonstrate not only the generalizability of the stacked ensemble model but also the discovery of three major categories of influencing factors based on forecasting method.Besides,the influencing modes of various factors on the air quality forecasting are elaborated,and the interpretability of forecasting method is enhanced.It helps to thoroughly cognize and understand the formation mechanism of serious haze events.