DDGNet: A Dual-Stage Dynamic Spatio-Temporal Graph Network for PM2.5 Forecasting

Dong Li, Haomin Yu, Yangli-ao Geng, Xiaobao Li, Qingyong Li

Abstract: As air pollution problems become increasingly serious, PM <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</inf> forecasting based on spatio-temporal observation data has received widespread attention. This forecasting task is full of challenges given the complicated producing factors and fickle transmission process of PM <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</inf> . However, most existing forecasting methods only exploit the spatial dependency by graph networks with fixed adjacency matrices, ignoring the dynamic spatio-temporal correlation of PM <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</inf> concentrations. In this paper, we propose a dual-stage dynamic spatio-temporal graph network (DDGNet) to model dynamic correlations for PM <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</inf> prediction of different cities. Specifically, DDGNet consists of two major stages: (1) dynamic graph construction to identify potentially informative neighbors for each node (a city) in every forecasting period; (2) graph attention networks to dynamically determine linking weights for each vertex to its neighbors. We evaluate DDGNet on three real-world datasets and compare it with several baselines. The experimental results demonstrate that our method achieves the state-of-the-art performance.