Prediction of residential indoor PM<sub>2.5</sub> concentration and model optimization based on Informer model

SUN Wen; YAN Weidong; HUANG Kailiang; SONG Jiasen

doi:10.13205/j.hjgc.202512018

Volume 43 Issue 12

Dec. 2025

Turn off MathJax

Article Contents

Article Navigation > ENVIRONMENTAL ENGINEERING > 2025 > 43(12): 161-168

SUN Wen, YAN Weidong, HUANG Kailiang, SONG Jiasen. Prediction of residential indoor PM2.5 concentration and model optimization based on Informer model[J]. ENVIRONMENTAL ENGINEERING , 2025, 43(12): 161-168. doi: 10.13205/j.hjgc.202512018

Citation:

SUN Wen, YAN Weidong, HUANG Kailiang, SONG Jiasen. Prediction of residential indoor PM_2.5 concentration and model optimization based on Informer model[J]. ENVIRONMENTAL ENGINEERING , 2025, 43(12): 161-168. doi: 10.13205/j.hjgc.202512018

Citation:

PDF( 1482 KB)

Prediction of residential indoor PM_2.5 concentration and model optimization based on Informer model

doi: 10.13205/j.hjgc.202512018

1. School of Municipal and Environmental Engineering, Shenyang Jianzhu University, Shenyang 110168, China;
2. School of Civil Engineering, Shenyang Jianzhu University, Shenyang 110168, China;
3. School of Civil Engineering, Chongqing University, Chongqing 400044, China

Received Date: 2024-12-05
Accepted Date: 2025-01-28
Rev Recd Date: 2025-01-10

Available Online: 2026-01-09

Abstract

Abstract

During the heating season in severe cold region， there is often a high concentration of PM_2.5 pollution indoors. Accurately predicting indoor PM_2.5 pollution levels is crucial for developing effective purification measures. The indoor PM_2.5 concentrations and window-opening behaviors of 7 residential buildings in a severe cold region during the heating season were analyzed. The PM_2.5 concentration exceeded the standard level by 35.94% in the living room， and the daily average duration of window-opening was relatively short. At the same time， the Informer model was trained using indoor PM_2.5 concentration historical data， with the duration of window-opening as a feature input to achieve the prediction of indoor PM_2.5 concentration. By comparing the prediction performance of the model under different feature input strategies， input step sizes， prediction ranges， and adding rolling prediction， the optimal configuration of the model to achieve high-precision prediction was determined after comprehensive evaluation. The model was capable of predicting indoor PM_2.5 concentrations for a future period， and the evaluation metrics for its prediction performance were： MAE of 10.70 μg/m³， RMSE of 13.75 μg/m³， and R² of 0.795. Compared to the TCN-LSTM model， the R² of the Informer model increased by 13.9%. The optimized model presents a more accurate prediction effect.
- indoor PM_2.5 pollution,
- PM_{2. 5} concentration prediction,
- Informer model,
- model optimization

FullText(HTML)

References(16)

References

[1]	HU W，DOWNWARD G S，REISS B，et al. Personal and indoor PM_2.5 exposure from burning solid fuels in vented and unvented stoves in a rural region of China with a high incidence of lung cancer［J］. Environmental Science and Technology，2014，48（15）：8456-8464.
[2]	SUNDELL J. On the history of indoor air quality and health［J］. Indoor Air，2004，14（S7）：51-58.
[3]	MA Y. Research on air quality prediction based on deep learning［D］. Ji'nan：Shandong University，2023. 马园．基于深度学习的空气质量预测研究［D］. 济南：山东大学，2023.
[4]	ZHU Y M，XU A L，SUN Qiang. New progress in air quality forecasting methods based on deep learning［J］. Environmental Monitoring in China，2020，36（3）：10-18. 朱晏民，徐爱兰，孙强．基于深度学习的空气质量预报方法新进展［J］. 中国环境监测，2020，36（3）：10-18.
[5]	DAI X L. Research on data mining and intelligent prediction of residential indoor air quality based on internet of things platform［D］. Tianjin：Tianjin University，2021. 戴希磊．基于物联网平台的住宅室内空气质量数据挖掘及其智能预测研究［D］. 天津：天津大学，2021.
[6]	ELBAYOUMI M，RAMLI N A，YUSOF N F F M. Development and comparison of regression models and feedforward backpropagation neural network models to predict seasonal indoor PM_2.5-10 and PM_2.5 concentrations in naturally ventilated schools［J］. Atmospheric Pollution Research，2015，6（6）：1013-1023.
[7]	LOY-BENITEZ J，VILELA P，LI Q，et al. Sequential prediction of quantitative health risk assessment for the fine particulate matter in an underground facility using deep recurrent neural networks［J］. Ecotoxicology and Environmental Safety，2019，169：316-324.
[8]	ZHANG N，BAI Z X，ZHANG H L，et al. Deployment，optimization and data prediction of Internet of Things for indoor air quality［J］. Sci China. Inf Sci，2022，52（1）：160-175. 张楠，白子轩，张泓亮，等．室内空气质量物联网部署、优化和数据预测［J］. 中国科学：信息科学，2022，52（1）：160-175.
[9]	DONG H，SUN L，OUYANG F. PM_2.5 concentration prediction based on Informer［J］. Environmental Engineering，2022，40（6）：48-54. 董浩，孙琳，欧阳峰. 基于Informer的PM_2.5浓度预测［J］. 环境工程，2022，40（6）：48-54.
[10]	HUANG K L，SUN W，FENG G H，et al. Indoor air quality analysis of 8 mechanically ventilated residential buildings in northeast China based on long-term monitoring［J］. Sustainable Cities and Society，2020，54：101947.
[11]	ZHOU H Y，LI J X，ZHANG S H，et al. Expanding the prediction capacity in long sequence time-series forecasting［J］. Artificial Intelligence，2023，318：103886.
[12]	CLEVERT D A，UNTERTHINER T，HOCHREITER S. Fast and accurate deep network learning by exponential linear units（ELUs）［EB/OL］.（2015-11-24）［ 2025-12-12］. arXiv：1511.07289［cs.LG］.
[13]	KINGMA D P，BA J. Adam：A method for stochastic optimization［EB/OL］.（2014-12-22）［ 2025-12-12］. arXiv：1412.6980［cs.LG］.
[14]	REN Y，WANG S Y，XIA B S. Deep learning coupled model based on TCN-LSTM for particulate matter concentration prediction［J］. Atmospheric Pollution Research，2023，14（4）：101703.
[15]	HUANG K L，SONG J S，FENG G H，et al. Indoor air quality analysis of residential buildings in northeast China based on field measurements and longtime monitoring［J］. Building and Environment，2018，144：171-183.
[16]	State Administration for Market Regulation，Standardization Administration of the People's Republic of China. GB/T 18883-2022：Indoor air quality standard［S］. Beijing：China Standards Press，2022. 国家市场监督管理总局，国家标准化管理委员会． GB/T 18883—2022：室内空气质量标准［S］. 北京：中国标准出版社，2022.