Regional Air Quality Meteorological Correlation Model Based On Industrial Waste Gas Emissions

Urban air pollution is an important problem that must be solved in the process of urban development.It not only has a negative impact on the stable development of social economy,but also threatens human health.With the acceleration of industrialization in China,the emission of pollutants from industrial enterprises is one of the main causes of urban air pollution.Under adverse meteorological conditions,the situation of air pollution is more serious.Ambient air quality prediction is an important part of air quality management and air pollution prevention and control.At present,there are two kinds of prediction methods of air quality:statistical method and machine learning algorithm.Due to the characteristics of high-dimensional non-linearity of the data used in the prediction,the traditional statistical methods to predict air quality have disadvantages such as low prediction accuracy and weak generalization ability.Air quality prediction based on machine learning algorithm can not only deal with high-dimensional nonlinear data,but also has the advantages of high prediction accuracy,generalization ability and strong anti-interference ability.However,current studies have separated themselves from pollutant emission information,or ignored pollutant source information,or used pollutant source inventory data to treat it as a constant.The working time of each process in the production enterprise is different,and the pollution load is not balanced,which leads to the fluctuation of the daily emissions of industrial waste gas.In view of this,three machine learning algorithms of PSO-LSSVM,random forest and GA-BP neural network were adopted in this study,and meteorological correlation models of air quality were established respectively based on the exhaust gas emissions and meteorological factors of industrial enterprises in the region.Combined with the prediction results of the models,regulation schemes for pollutant emissions of industrial enterprises in the study area were provided.First,this study is based on AQI and six atmospheric pollutant concentration data of PM_2.5、PM₁₀、NO₂、SO₂、CO and O₃ on January 1,2016 and December 31,2018 in Zhangdian District,Zibo City.According to the principal component analysis,the main air pollutants affecting the air quality in Zhangdian District are PM_2.5、PM₁₀、NO₂、SO₂、CO and O₃ in order from large to small.Then,the AQI data of January 1,2018 solstice and October 14,2018 in Zhangdian District,meteorological factors and daily industrial waste gas emissions were used as the training samples of the AQI prediction model.The AQI data on October 15,2018 solstice and November 3,meteorological factors and industrial waste gas daily emissions were used as test samples of the AQI prediction model.Using MATLAB,Python and other software,three machine learning algorithms of PSO-LSSVM,Random forest and GA-BP neural network were used to establish the meteorological correlation model of air quality.The predicted AQI values of the three models were almost exactly the same as the measured values in the time series trend,among which the air quality meteorological correlation model based on PSO-LSSVM algorithm had the best prediction performance,with RMSE,MAE and R²of 9.8431,7.5037 and 0.8950 respectively.Finally,the PSO-LSSVM air quality meteorological correlation model was used to retrieve the daily emission limit of industrial waste gas on October 15 solstice and November3 in Zhangdian District,and on the basis of ensuring good environmental air quality,the industrial enterprises in this area were tried to guide the reasonable adjustment of their production process.This research method can provide reference for urban air quality protection and pollution emission control of industrial enterprises.