A Novel Prediction Approach for Exploring PM2.5 Spatiotemporal Propagation Based on Convolutional Recursive Neural Networks

The spread of PM2.5 pollutants that endanger health is difficult to predict because it involves many atmospheric variables. These micron particles can spread rapidly from their source to residential areas, increasing the risk of respiratory disease if exposed for long periods. The prediction system of PM2.5 propagation provides more detailed and accurate information as an early warning system to reduce health impacts on the community. According to the idea of transformative computing, the approach we propose in this paper allows computation on the dataset obtained from massive-scale PM2.5 sensor nodes via wireless sensor network. In the scheme, the deep learning model is implemented on the server nodes to extract spatiotemporal features on these datasets. This research was conducted by using dataset of air quality monitoring systems in Taiwan. This study presents a new model based on the convolutional recursive neural network to generate the prediction map. In general, the model is able to provide accurate predictive results by considering the bonds among measurement nodes in both spatially and temporally. Therefore, the particulate pollutant propagation of PM2.5 could be precisely monitored by using the model we propose in this paper.