Prediction And Mitigation Strategies Of Air Pollutants In The Yangtze River Delta Region Based On Novel Grey Nonlinear Models

In the context of acceleration of industrialization and urbanization in China,some urban clusters are suffering severe atmospheric pollution,such as the Yangtze River Delta.Accurate air quality prediction plays a key role for environmental protection departments in designing the early-warning system of air pollution prevention and control,tailoring relevant policy-decisions and realizing the sustainable development of economy and society.Since the seasonal variation in meteorological circumstances causes the seasonal fluctuation of air quality and various pollutant index series to a large extent,it is difficult to describe and extract by using traditional forecasting tools.To solve this problem,this paper developes two novel nonlinear seasonal gray models,abbreviated as SNGBM（1,1）and SWBGM（1,1）,on the basis of the traditional nonlinear grey models,which can effectively capture the seasonal and non-linear characteristics of the original data.And they were utilized to model air quality indicators in the long term（quarterly）and PM_2.5concentration in the short term（monthly）in four major cities in the Yangtze River Delta region,respectively.In terms of the features of sparsity,finiteness and uncertainty in air quality system of China,it is difficult to achieve accurate prediction by using traditional predictive tools.Considering that grey system theory aimed at solving problems of few data,poor information and uncertain systems,this paper applies grey predictive models to study air quality indicators.Based on the traditional nonlinear grey models NGBM（1,1）and WBGM（1,1）models,the seasonal adjustment factors which can effectively capture the characteristics of seasonal changes in time series are introduced into them respectively,thereby constructing the nonlinear seasonal grey models SNGBM（1,1）and SWBGM（1,1）.And the intelligent optimization legal culture algorithm is selected to optimize the super parameters of the new models.For the purpose of evaluating the simulative and predictive performance of the new models,this paper applied SNGBM（1,1）for forecasting quarterly air quality indicators,while SWBGM（1,1）for forecasting monthly PM_2.5concentration in Shanghai,Nanjing,Hangzhou and Hefei respectively,then compared theirs predictive performance with comparison models in terms of several accuracy evaluation criteria.The empirical results show that both new models have better simulation and prediction accuracy and stronger prediction stability,suggesting that the new models could dispict the seasonal periodicity and nonlinear fluctuation characteristics of time series,compared with other gray models,statistical econometric models and machine learning models.In addition,air quality in Shanghai,Hangzhou and Hefei will improve in the next two years,while that in Nanjing will decline slightly,according to the forecasted results.Based of the forecasting results of air quality in the YRD region,this paper proposes the targeted policy suggestions to achieve prevention and control of air pollution.Given the drawbacks of the nonlinear seasonal grey prediction model,future research direction is poinded out.That is,to further enhance the simulation and prediction accuracy of the grey prediction model,a new multivariable nonlinear seasonal grey model by taking various influencing factors into consideration can be developed in the future.