A Novel Predictor for Exploring PM2.5 Spatiotemporal Propagation by Using Convolutional Recursive Neural Networks

A Novel Predictor for Exploring PM2.5 Spatiotemporal Propagation by Using Convolutional Recursive Neural Networks

The spread of PM2.5 pollutants that endanger health is difficult to predict because it involves many atmospheric variables. These micro particles could spread rapidly from their source to residential areas, increasing the risk of respiratory disease if exposed for long periods. However, the existing prediction systems do not take into account the geographical correlation among neighboring nodes spatially and temporally resulting in loss of important information, lack of PM2.5 propagation resolution, and lower forecasting accuracy. In this paper, a novel scheme is proposed to generate propagation heat maps of PM2.5 prediction by using spatiotemporal datasets. In this scheme, the deep learning model is implemented to extract spatiotemporal features on these datasets. This research was conducted by using the dataset of air quality monitoring systems in Taiwan. Moreover, the robust model based on the convolutional recursive neural network is presented to generate the propagation maps of PM2.5 concentration. This study develops an intelligence-based predictor by using Convolutional Recursive Neural Network (CRNN) model for predicting the PM2.5 propagation with uncertain spread and density. It is also one of key technologies of software and hardware co-design for massive Internet of Things (IoT) applications. Finally, the proposed model the proposed model provides accurate predictive results over time by taking into account the spatiotemporal relationship among sensory nodes in order to give a prediction solution for the massive IoT deployment based on green communication.