Adaptive energy management of Swedish residential buildings using artificial intelligence and reinforcement learning for enhanced climate resilience

As the impact of climate change intensifies, there is an urge to adapt Swedish residential buildings with a special focus on improving their energy efficiency and resilience. In this study, adaptive strategies through Artificial Intelligence and Reinforcement Learning will be provided for the energy management of different Swedish residential buildings. The aim will be the optimization of energy use in future climate scenarios particularly during extreme weather conditions of heat waves or cold snaps.

Previous research has proven the potential of AI and RL for dynamic energy management. RL-based approaches have already achieved considerable energy saving and improved climate resilience in applications grid-connected under severe weather conditions. Research into their applicability in residential buildings under these conditions remains relatively unexplored.

In view of this, the objective of this research is to close this gap by developing and evaluating robust predictive models for buildings of various types and sizes, comparing their performance against predicted future climate scenarios. This includes the possibility of applying the RL and AI algorithms for adaptive management of building systems, starting from set-point adjustments to changes in air ventilation rate.

Further, dynamic simulations monitor and regulate the performance of buildings in real-time. Simulation-based control enables the study to vary strategies of control with updated data in real-time for reduced energy use and improved indoor comfort despite climatic changes outdoors.

Preliminary results show that AI- and RL-based strategies, when combined with dynamic simulations, can improve energy efficiency and resilience to extreme weather events. The simulations introduce a detailed degree of control, which allows for very fine adjustment for external dynamic variations. This achieves indoor optimal conditions while saving energy by avoiding unnecessary consumption during mild conditions and enhancing its efficiency in the case of extreme events.

The results of this study should help in the development of smarter and more sustainable technologies for buildings and should provide important insights into climate adaptation within urban planning. By implementing AI, RL, and dynamic simulations in energy management, this research is assisting in shifting toward climate-resilient residential buildings and supporting additional goals about energy sustainability and climate action.