Adaptive optimal control scheme of HVAC system based on supervised learning algorithm

Proposes a real-time adaptive control scheme of applying supervised learning algorithms to the control of HVAC systems. Comparing with the semi-physical model optimal control, the proposed method can make use of simple machine learning models and be automatically updated online, so as to adapt to system degradation and/or sensor errors. Conducts the dynamic validation tests for the cooling tower system in a high-rise building. The results show that the proposed scheme has significant advantages over the semi-physical model based on optimal control method.