Adaptive and Predictive Control Method to Improve Energy Efficiency and Indoor Thermal Comfort for Intermittently Occupied Spaces

Purpose: Beyond the essential concerns of economy and maintenance for intermittently occupied spaces, there’s a critical need to investigate high performance control methodologies. The goal is a dual imperative: significantly enhancing occupant thermal comfort while optimizing the energy performance profile of these often-fleeting architectural assets. Method: The core method integrates a conventional thermostat model with an adaptive control process. This adaptive layer continuously analyzes and incorporates real-time changes in human-centric factors (like occupancy and adaptive comfort expectations) as dynamic system inputs. These refined control patterns, which integrate both the building’s physics and occupants’ behavior, are used to train a machine learning-based predictive control model. The subsequent analysis rigorously assesses the model’s control validity and energy efficiency. Results: The performance assessment demonstrates significant efficiency for the developed learning predictive model when compared to the baseline conventional thermostat control. In thermal comfort consistency, a substantial improvement of approximately 84.8% is observed in meeting the thermal comfort index. In operational energy reduction, the model achieves an estimated 7.4% reduction.