Energy Management Agent for Regulating Particulate Matter in Railway Stations with Photovoltaic Power

This study addresses the challenge of managing fine dust (PM2.5 and PM10) levels in underground train stations, where air quality is compromised due to limited ventilation and various pollution sources. Traditional methods struggle to optimize control systems for dust reduction, particularly when accounting for station-specific variables like depth and congestion. To address this, the study proposes a machine learning-based energy management agent using a Deep Q-Network (DQN) integrated with an artificial neural network (ANN). The ANN predicts dust concentration changes based on fan and air conditioning controls, while the DQN optimizes these controls to balance dust reduction and energy costs. Additionally, the model considers the integration of photovoltaic power to enhance energy efficiency. The approach was validated using data from Namgwangju Station, demonstrating improved air quality and energy efficiency.