Multi-hour and multi-site air quality index forecasting in Beijing using CNN, LSTM, CNN-LSTM, and spatiotemporal clustering

作者：

Highlights：

• The AQI forecasting model uses deep learning and spatiotemporal clustering.

• The multiple-site forecasting models were developed for the next 1–6 h.

• The overall forecasting for all the stations in Beijing through LSTM is optimal.

• Seasonal or spatial clustering-based forecasting is suitable for a season or cluster.

• CNN-LSTM and LSTM generally outperform CNN and BPNN.

摘要

•The AQI forecasting model uses deep learning and spatiotemporal clustering.•The multiple-site forecasting models were developed for the next 1–6 h.•The overall forecasting for all the stations in Beijing through LSTM is optimal.•Seasonal or spatial clustering-based forecasting is suitable for a season or cluster.•CNN-LSTM and LSTM generally outperform CNN and BPNN.

论文关键词：LSTM,CNN,Forecasting,AQI,Spatiotemporal clustering

论文评审过程：Received 8 September 2020, Revised 12 November 2020, Accepted 11 December 2020, Available online 24 December 2020, Version of Record 1 January 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.114513