Prediction of Local PM2.5 Concentrations Based on Time-Series Feature Learning Using Multivariate LSTM

Air pollution causes various health problems and diseases. Long-term PM2.5 monitoring and prediction of its occurrence and sources are necessary not only in global areas based on public monitoring stations but also in local areas using cost-effective sensor systems. For this study, we developed a sensor system to achieve simplified and high-frequency PM2.5 measurements. We attempted to learn and to predict local PM2.5 concentrations from observed data using long short-term memory (LSTM) as a dominant time-series feature learning network. For improving learning and prediction accuracy evaluated according to the root mean square error (RMSE), sensor calibration is performed using a higher sensor. Moreover, we strove to reduce RMSE by optimizing its five major parameters. Experimentally obtained results demonstrate that the prediction accuracy is improved gradually after calibration and parameter optimization. As an ablation experiment, five meteorological factors are imported externally to verify the factors which contribute to reducing RMSE. Results verify the strong effects of local pressure and temperature for training and relative humidity and temperature for testing as validation.

1. Introduction

Air pollution is caused predominantly by chemical substances and gases emitted mainly from power stations, factories, automobiles, and airplanes. Freely floating air pollutants in the atmosphere can pose various health hazards to humans and to animals and plants of many kinds. Particulate matter (PM) is a highly impactful air pollutant. Actually, PM_2.5 with particle size less than 2.5 increase the risk of respiratory diseases such as asthma and bronchitis because they tend to penetrate deeply into the lungs. This increased risk is also a concern as a cause of lung cancer and its effects on the cardiovascular system. According to research by Liu et al. [1], 0.55% of the cardiovascular mortality, 0.74% of the respiratory mortality, and 0.68% of the total mortality rate for all causes of death are increased if the two-day moving average of PM_2.5 concentrations increases by 10 .