Nonlinearity in thermal comfort-based control systems: A systematic review

Abstract

• Exploring fundamental research on thermal comfort nonlinearity and evaluation approaches. • Analysing adaptive thermal comfort-based control for space cooling. • Emphasizing differences linearisation and model reduction in thermal feedback control. • Suggesting future directions for thermal comfort control, and system identification. This work presents an in-depth systematic literature review of the strategies used to characterise and quantify thermal comfort conditioned by mechanical HVAC systems. The model development is paramount for the study on the stability and robustness of the ventilation process control. They are required to establish supervisory and local control loops to improve the component sequence of the HVAC systems, and the interaction among the indoor environment, occupants, and the HVAC system response. Over the past decade, innovative technologies and artificial intelligence revamped the HVAC control research with a reluctancy of application in practice due to complex computations and lack of understanding for these innovations. However, the need to find the balance between the functionality of the HVAC system and suitable comfort levels of occupants still persisted. This work examines three research clusters of HVAC systems: zone thermal conditions representation, the inherent nonlinearity of HVAC complex systems and model reduction strategies, and HVAC processes optimisation and control methods . This enables a holistic view of the complexities folded in delivering occupants’ thermal comfort. Central to constructing these research clusters, existing studies were investigated following the four-step protocol based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines: identification of relevant literature based on keywords, screening of the chosen literature quality, setting eligibility criteria for scoping the objectives of the study and lastly inclusion of literature categories into one set. Based on the findings of this work, the application of linearisation and model reduction is found to be a promising area of research in the field of HVAC system control and optimisation. This is because emerging deep learning methods could facilitate the integration of linearisation to high-dimensional data of zone thermal interactions presented by computational fluid dynamics. Finally, this work discusses the challenges faced in the data-driven control strategies of HVAC systems, opens future direction of evaluating occupants’ comfort and highlights the importance of interpretability and tracking of control process in relation to thermal comfort.

Topics & Concepts