Optimization of Energy-Efficient HVAC Systems in Bio-Safe Environments using Reinforcement Learning for Predictive Maintenance and Real-Time Control

Abstract

Bio-safety compliance and energy efficiency are still a pressing issue in the management of modern buildings, especially in such a setting as a healthcare facility or a laboratory, where the accuracy of controlling the environmental conditions is the most important factor. Reinforcement Learning (RL) is potentially a solution to the optimization of HVAC performance by dynamically modifying operational parameters (e.g. temperature, humidity, airflow, and energy consumption) to achieve a more efficient HVAC system. The paper introduces a RL-based solution to optimize HVAC systems in bio-safe settings, which uses real-time information provided by Internet of Things-based sensors to constantly monitor and modify system settings. Trial and error allow the RL agent to learn how to make the best decisions, enhance the energy efficiency, but still meet high bio-safety standards. The proposed system simulation was carried out in MATLAB, and the RL-based HVAC system demonstrated an average consumption of 17% of energy, along with ensuring bio-safety requirements in sensitive areas. It enabled the system to anticipate and thwart HVAC breakdowns, saving 25% of downtime and improving the overall reliability of the system. This dynamic optimization model shows that RL has the potential to develop sustainable and cost-efficient solutions to HVAC systems, which can have long-term benefits in terms of energy management and occupant safety.