Review of HVAC forecasting and control strategies for improved building performance

Abstract

Efficient Heating, Ventilation, and Air Conditioning (HVAC) operation strategies are critical to achieving both high energy efficiency and satisfactory Indoor Environmental Quality (IEQ) in modern buildings. Recently, data-driven strategies, particularly those leveraging Artificial Intelligence (AI) techniques, have demonstrated superior forecasting accuracy and adaptive control capabilities, enabling more responsive and efficient HVAC system management. This study summarises recent forecasting and control strategies for building HVAC systems, categorising them into three research perspectives namely, improving IEQ for occupants, enhancing HVAC energy efficiency for building managers, and intelligent methods in control and environmental analysis to support innovation for researchers. While data-driven methods demonstrate higher precision and flexibility, challenges remain regarding the dataset quality, uncertainty handling, computational cost, sim-real validation, and model interpretation. Hybrid methods that integrate model-based and data-driven techniques, along with advanced feature fusion and extraction, show strong potential to improve robustness, transparency, and generalisation of strategies in condition forecasting and building HVAC control. Furthermore, this study also finds that considering different time-scale information is a valuable direction to further enhance HVAC predictive control performance. Tackling these gaps will be pivotal for accelerating the deployment of next-generation HVAC control systems that can effectively balance energy efficiency and occupant comfort in large buildings.