Abstract
This paper presents the implementation and analysisof a ML-based MPC system integrated with a real-timeoccupant feedback mechanism through field tests in acommercial building. We demonstrate the efficacy of theoccupant-centric control strategy that dynamicallyadjusts thermal comfort setpoints based on direct userinput. Furthermore, we propose a comprehensiveframework for analyzing the resulting data, offeringinsights into system performance, energy consumption,and occupant satisfaction. The findings illustrate thatincorporating a human-in-the-loop approach canenhance building energy efficiency withoutcompromising occupant comfort.
Keywords Model Predictive Control, Machine Learning,AI, PMV, ACMV, Occupant Feedback
Copyright ©
Energy Proceedings