Deep Air Quality Forecasting Using Hybrid Deep Learning Framework

Air quality forecasting has been regarded as the key problem of air pollution early warning and control management. In this article, we propose a novel deep learning model for (mainly PM2.5) forecasting, which learns spatial-temporal correlation features interdependence multivariate related time series data by hybrid architecture. Due to nonlinear dynamic characteristics data, base modules our include one-dimensional Convolutional Neural Networks (1D-CNNs) Bi-directional Long Short-term Memory networks (Bi-LSTM). The former is extract local trend spatial features, latter learn dependencies. Then design jointly framework based on CNNs Bi-LSTM shared representation data. We conduct extensive experimental evaluations using two real-world datasets, results show that capable dealing with PM2.5 satisfied accuracy.