Scholarlink

Purpose <p>The aim of this study was to investigate the impact of improved deep learning model on the predictive performance of PM<sub>2.5</sub> concentration.</p> Methods <p>We developed a new model combining one-dimensional convolutional neural network and bidirectional long short-term memory neural network to predict PM<sub>2.5</sub> concentrations at hourly intervals. The air pollution observation data from 2020 to 2022 collected at several national air quality monitoring stations in Shenyang (Liaoning province, China) were employed to train our model. The performance of the proposed model was boosted by connecting the layer of network calculated results with the PM<sub>2.5</sub> sequence data. Furthermore, data of most relevant air quality monitoring stations and PM<sub>2.5</sub> feature factors of the target station were screened. The spatial correlation of major air pollutant and the interaction between PM<sub>2.5</sub> and other pollutant factors were therefore considered to improve the accuracy of the model.</p> Results <p>The root mean square error, mean absolute error, mean absolute percentage error of the new method were reduced by 49%, 51%, 44% and the R<sup>2</sup> was improved by 4.6% respectively compared with the control group for the next hour prediction. The proposed improvement method can reduce the prediction error of the model in the next 6&#xa0;h.</p> Conclusions <p>In this study, the proposed model improvement method can significantly reduce the error of the model in predicting PM<sub>2.5</sub> concentration. The proposed method can improve the model in the next 6&#xa0;h prediction accuracy. This study provides a new perspective for establishing high-precision models for PM<sub>2.5</sub> prediction.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhanced short-term prediction of urban PM2.5 concentrations by improved hybrid deep learning

  • 期刊论文

  • 发表日期：

    2025年8月25日

查看全文
  • Yin Zhou,
  • Yaping Lyu,
  • Xiuli Dang,
  • Roland Bol,
  • Peng Zhang,
  • Na Yu,
  • Yuling Zhang

摘要

Purpose

The aim of this study was to investigate the impact of improved deep learning model on the predictive performance of PM2.5 concentration.

Methods

We developed a new model combining one-dimensional convolutional neural network and bidirectional long short-term memory neural network to predict PM2.5 concentrations at hourly intervals. The air pollution observation data from 2020 to 2022 collected at several national air quality monitoring stations in Shenyang (Liaoning province, China) were employed to train our model. The performance of the proposed model was boosted by connecting the layer of network calculated results with the PM2.5 sequence data. Furthermore, data of most relevant air quality monitoring stations and PM2.5 feature factors of the target station were screened. The spatial correlation of major air pollutant and the interaction between PM2.5 and other pollutant factors were therefore considered to improve the accuracy of the model.

Results

The root mean square error, mean absolute error, mean absolute percentage error of the new method were reduced by 49%, 51%, 44% and the R2 was improved by 4.6% respectively compared with the control group for the next hour prediction. The proposed improvement method can reduce the prediction error of the model in the next 6 h.

Conclusions

In this study, the proposed model improvement method can significantly reduce the error of the model in predicting PM2.5 concentration. The proposed method can improve the model in the next 6 h prediction accuracy. This study provides a new perspective for establishing high-precision models for PM2.5 prediction.