Air Quality Forecast Of Graph Convolutional Neural Network Based On Fusion Of Spatial Information

In recent years,with the rapid development of urbanization,the problem of air pollution has become increasingly serious,which has greatly affected people’s daily life and physical health.The Air Quality Index(AQI) is a numerical evaluation index for evaluating air quality.It combines the concentration of several air pollutants routinely monitored by environmental quality standards and the impact of various pollutants on human health,ecology and the environment,so as to intuitively reflect the degree of urban air pollution.Therefore,in order to theoretically guide the control of air pollution and improve the air quality of residents,it is necessary to establish a model that can accurately predict the future air quality index.In this paper,the air quality index in Shanghai research and prediction map convolution neural network model based on the integration of spatial information.First of all,this article establishes the graph network of Shanghai air quality monitoring station based on the spatial correlation of air pollutants.Then,according to the distance relationship between each air quality monitoring station,the adjacency matrix of the graph is calculated,and then the spatial location adjacency of each air quality monitoring station and the similarity of air quality data are considered,and finally the characteristics between airspace data are extracted.In addition,this article also analyzes the dynamic behavior of time series data through the linear convolution module of the gating unit.Subsequently,on the basis of LSTM and CNN,this paper uses the combination of time-gated and space-gated blocks to construct the basic structure of the prediction model,and adds the graph attention network structure to the depth model to further optimize the model.On this basis,the spatial and temporal data are fused to obtain a new fusion model.Finally,this paper compares the prediction results of the LSTM model,the TGCN model,and the spatiotemporal model that combines spatial information.The method proposed in this paper not only considers the time correlation between air pollutants in each air quality monitoring site,but also combines the effects of surrounding monitoring sites on air quality index prediction.The experimental results show that compared with other models,the method proposed in this paper has a great improvement in the accuracy of the results and the generalization ability of the model.