Predicting PM_2.5 in Well-Mixed Indoor Air for a Large Office Building Using Regression and Artificial Neural Network Models

Abstract

Although the exposure to PM 2.5 has serious health implications, indoor PM 2.5 monitoring is not a widely applied practice. Regulations on the indoor PM 2.5 level and measurement schemes are not well established. Compared to other indoor settings, PM 2.5 prediction models for large office buildings are particularly lacking. In response to these challenges, statistical models were developed in this paper to predict the PM 2.5 concentration in well-mixed indoor air in a commercial office building. The performances of different modeling methods, including multiple linear regression (MLR), partial least squares regression (PLS), distributed lag model (DLM), least absolute shrinkage selector operator (LASSO), simple artificial neural networks (ANN), and long-short term memory (LSTM), were compared. Various combinations of environmental and meteorological parameters were used as predictors. The root-mean-square error (RMSE) of the predicted hourly PM 2.5 was 1.73 μg/m 3 for the LSTM model and in the range of 2.20−4.71 μg/m 3 for the other models when regulatory ambient PM 2.5 data were used as predictors. The LSTM models outperformed other modeling approaches across the performance metrics used by learning the predictors' temporal patterns. Even without any ambient PM 2.5 information, the developed models still demonstrated relatively high skill in predicting the PM 2.5 levels in well-mixed indoor air.

Atmospheric Deposition of Particulate Matter and Micropollutants as a Major Mass Transport Route to Surface Water in Winter: Measurement and Modeling in Beijing in 2014 and 2021

“…Model performance was monitored using two indicators: the Pearson correlation coefficient ( R , which evaluates the model fitness between the measured and simulated values) and the root-mean-square error (RMSE, which represents the global error between the measured and simulated values). ,,− Assuming that there are n groups of the measured values and simulated values, the equations of R and RMSE are as follows where Pred i is the i -th common predicted value, is the average of all predicted values, Real i is the i -th actual measured value, and is the average of all measured values. The range of R and MSE is [1̅,1] and [0, + ∞], respectively.…”

“…The closer the R value is to 1, the closer the MSE value is to 0, indicating that the model is more reliable. A perfect fitting will result in R = ± 1 and MSE = 0 …”

“…Model performance was monitored using two indicators: the Pearson correlation coefficient (R, which evaluates the model fitness between the measured and simulated values) and the root-mean-square error (RMSE, which represents the global error between the measured and simulated values). 31,34,39−41 Assuming that there are n groups of the measured values and simulated values, the equations of R and RMSE are as follows 31…”

“…Herein, we employed a neural network (NN) model to predict the impact of heavy metals and PAHs in the atmospheric fine particles on urban surface water quality. NN has been used to predict atmospheric nitrogen deposition in aquatic ecosystems, the concentration of heavy metals in PM 1 , PM 2.5 concentration, NO 2 concentration, O 3 concentration, water quality, − etc. However, few studies have focused on using atmospheric pollutants and meteorological factors to statistically characterize heavy metals and PAHs in surface waters.…”

AI-big data analytics for building automation and management systems: a survey, actual challenges and future perspectives

“…In a similar way, in Lagesse et al. ( 2020 ), various ML models are utilized for predicting in office buildings, i.e. ANN, LSTM, multiple linear regression (MLR), partial least squares regression (PLS), distributed lag model (DLM), and least absolute shrinkage selector operator (LASSO).…”