Title: A CA-GRU-based model for air quality prediction

Authors: Jingyang Wang; Xiaolei Li; Tingting Wang; Jiazheng Li; Qiuhong Sun

Addresses: School of Information Science and Engineering, Hebei University of Science and Technology, Shijiazhuang 050018, China ' School of Information Science and Engineering, Hebei University of Science and Technology, Shijiazhuang 050018, China ' School of Information Science and Engineering, Hebei University of Science and Technology, Shijiazhuang 050018, China ' School of Information Science and Engineering, Hebei University of Science and Technology, Shijiazhuang 050018, China ' School of Information Science and Engineering, Hebei University of Science and Technology, Shijiazhuang 050018, China

Abstract: Smart cities aim to maximise the optimisation of urban functions, promote economic growth, and use smart technology and data analysis to improve the life quality of urban residents. The air quality index (AQI) is an important evaluation index of air pollution, describing the degree of air pollution and its impact on the health of human beings. Therefore, it is particularly important for smart cities to accurately predict AQI. The accuracy of AQI prediction remains a challenge for current methods. In this paper, an AQI prediction model based on CA-GRU is proposed, which includes convolutional neural networks (CNN), attention mechanism, and gated recurrent unit (GRU). In this model, the feature extraction of input data is realised through CNN, weights are assigned according to the states of old data with attention mechanism and the AQI is predicted with GRU. To prove the validity and accuracy of the CA-GRU prediction model, we take experiments using air quality data and weather data from 1st January 2017 at 00:00 to 30th September 2020 at 23:00 in Shijiazhuang City of China and compare this model with other models. Finally, mean absolute error (MAE), mean square error (MSE), explained variance score (EVS), and R² are used to evaluate the performance of the model in this experiment. The results shows that this model has the highest performance than the others, with the MAE of 6.099281, MSE of 90.781522, EVS of 0.972560, and R² of 0.972495.

Keywords: convolutional neural network; CNN; attention; gated recurrent unit; GRU; air quality index; AQI; AQI prediction.

DOI: 10.1504/IJAHUC.2021.119098

International Journal of Ad Hoc and Ubiquitous Computing, 2021 Vol.38 No.1/2/3, pp.184 - 198

Received: 04 Mar 2021
Accepted: 31 Mar 2021
Published online: 22 Nov 2021 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article