TY - JOUR
TI - Forecasting indoor NO2 concentrations in a student dormitory using RNN-LSTM models
AU - Xie Jia
AU - Wu Banglong
AU - Yang Shuting
AU - Xia Yaowen
JN - Thermal Science
PY - 2026
VL - 30
IS - 2
SP - 1453
EP - 1465
PT - Article
AB - Air quality has a profound impact on the urban environment, with both positive and negative effects. Developing effective strategies for improving and predicting air quality is crucial for urban environmental management. This study was conducted from June 2021 to March 2022, with a sample size of 584. The study collected air quality data from dormitory buildings, including indoor temperature, wind speed, humidity, and NO2 concentration. Additionally, we conducted a qualitative analysis to explore the relationship between atmospheric parameters and average NO2 concentration. We compared multilayer perceptron with recurrent neural networks (RNN) and long short-term memory (LSTM) models for optimization in predicting NO2 concentrations. Experimental results showed that the combination of RNN and LSTM significantly improved the accuracy of NO2 concentration predictions. This study provides important references for monitoring NO2 concentrations in university dormitories and improving air quality.
DO - 10.2298/TSCI250602161X
ER -