Bayesian Optimization Framework for HVAC System Control

The use of machine-learning algorithms in optimizing the energy efficiency HVAC systems has been widely studied recent years. Previous research focused mainly on data-driven model predictive controls and reinforcement learning. Both approaches require a large amount online interactive data; therefore, they are not efficient stable enough for large-scale practical applications. In this paper, Bayesian optimization framework control proposed to achieve near-optimal performance while also maintaining high stability, which would allow it be implemented number projects obtain benefits. includes following: (1) method modeling problems as contexture technology automatically constructing samples, based time series raw trending (2) Gaussian process regression surrogate objective function optimization; (3) loop, optimized characteristics system controls, including an additional exploration trick noise estimation mechanism ensure constraint satisfaction. was evaluated by using simulation system, calibrated data from real center. results our study showed that approach achieved more than 10% increase energy-efficiency savings within few weeks compared with original building automation control.