Towards energy flexible commercial buildings: Machine learning approaches, implementation aspects, and future research directions

Abstract

Commercial buildings encounter considerable challenges in predicting and managing energy flexibility, arising from the complexity of their energy systems and the interdependencies among system components and building thermal mass. Nonetheless, the emergence of “smarter buildings” creates significant opportunities for applying machine learning (ML) techniques in energy flexibility. These methods provide significant benefits to commercial building owners, with multiple states integrating energy flexibility provisions for commercial buildings into their regulatory frameworks. This paper provides a systematic review of the role of commercial buildings in energy flexibility studies, with a particular emphasis on ML techniques used in the characterisation, optimisation, and forecasting of energy flexibility. Furthermore, it examines the direct monetary and non-direct monetary benefits and practical challenges associated with integrating flexibility concepts into commercial buildings, as well as the policy and regulatory frameworks that facilitate flexibility implementations. A comprehensive understanding of these aspects will be beneficial for developing robust frameworks that enhance the adaptive capacity of commercial buildings, thus enabling their seamless integration into dynamic energy markets while supporting grid stability and sustainability objectives.