Research Article
Prediction of the Short-Term PM2.5 Concentration Based on Informer
@INPROCEEDINGS{10.1007/978-3-031-65123-6_15, author={Jijing Cai and Chen Wang and Le Yu and Meilei Lv and Kai Fang}, title={Prediction of the Short-Term PM2.5 Concentration Based on Informer}, proceedings={Quality, Reliability, Security and Robustness in Heterogeneous Systems. 19th EAI International Conference, QShine 2023, Shenzhen, China, October 8 -- 9, 2023, Proceedings, Part II}, proceedings_a={QSHINE PART 2}, year={2024}, month={8}, keywords={PM2.5 forecast Informer Wiener filter Principal component analysis}, doi={10.1007/978-3-031-65123-6_15} }- Jijing Cai
Chen Wang
Le Yu
Meilei Lv
Kai Fang
Year: 2024
Prediction of the Short-Term PM2.5 Concentration Based on Informer
QSHINE PART 2
Springer
DOI: 10.1007/978-3-031-65123-6_15
Abstract
PM2.5 poses a significant health risk to individuals. Hence, precise prediction of PM2.5 concentration is of utmost importance. This article proposed a novel hybrid deep learning model called WPI for short-term PM2.5 prediction. The model utilizes wiener filtering and principal component analysis to denoise and extract features from meteorological data consisting of 10 variables. By combining these techniques with the Informer model, the WPI model achieves accurate short-term prediction of PM2.5 concentration. To evaluate the performance of the WPI model, three types of metrics are used: root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination ((R^2)). In short-term prediction, the WPI model achieves the(R^2)value at least 0.0418 higher than the comparison model, with RMSE at least 0.4647 lower and MAE at least 0.5507 lower.
