A proactive energy-efficient optimal ventilation system using artificial intelligent techniques under outdoor air quality conditions

Abstract Passengers are directly or indirectly exposed to fine dust in indoor air of underground subway stations, which greatly affects the comfort and health of the passengers. However, conventional ventilation systems are manually operated, resulting in high energy consumption, and it is hard to consider the dynamic characteristics of indoor air quality due to the complex relationships between the environment of the subway station and climate change-driven outdoor air quality. Therefore, an energy-efficient ventilation optimization system based on deep learning and artificial intelligence (AI)-iterative dynamic programming was developed in this study for proactive environmental and economic maintenance of the underground ventilation system for a subway’s indoor air quality. The deep learning model predicted the next 24 h of the subway’s environmental status, and the AI-iterative dynamic programming searched a piecewise operational policy of ventilation flow rate for the same operational duration. Energy efficiency was improved by 8.68% while maintaining healthy indoor air quality for the passengers. The proactive optimal ventilation system for the platform of a target subway station presented a decrease of 96 tons of CO2 per year to help address climate change and operating expenditure savings of up to $4217 dollars per year.