Regional Air Quality Modeling And Analysis Method Research Based On Complex Network

In recent years,while China has made great progress in industrialization and urbanization,Industrial civilization and urban development have also had a tremendous impact on the ecological environment.With the frequent outbreak of "haze fog",air quality issues have caused widespread concern.The air quality problems with inhalable particulate matter,sulfur dioxide,and nitrogen oxide compounds as main pollutants have become increasingly serious.These pollutants exposed to the atmosphere have also affected the sustainable development of China's economy while also affecting public health.Under this situation,this paper takes the air quality in the Beijing area as the research object,analyzes the influencing factors through the air quality monitoring site monitoring data,proposes the use of complex networks to extract the time and space influence factors,and proposes the use of time and The spatial prediction model combines methods for air quality prediction.First,the correlation between pollutant concentrations was calculated using Pearson's correlation coefficient,and the influence of meteorological factors and geographical factors on each pollutant was analyzed from the time and space dimensions.Then,aiming at the spatial heterogeneity of air quality data,a complex network was proposed to abstract each site where each pollutant is located to a node of a complex network,and at the same time,the concentration of each pollutant spreads in space to the edge of a complex network.The edge weights abstract the spatial interaction network and indicate the spatial impact of other sites on the predicted site.Secondly,using the extreme learning machine(Extreme Learning Machine,ELM)as input,meteorological data and spatial influence data to establish a spatial prediction model.Aiming at the temporal correlation of air quality data,a long-term and short-term memory network(Long and short term memory network,LSTM)model was proposed.The model was predicted based on historical air quality data.Finally,the collaborative training model was adopted to integrate the forecast results of time and space,and the prediction results of the three models were analyzed.It was verified that the spatio-temporal collaborative forecasting model was better than the spatial-temporal single forecasting model.