No abstract

No abstract

No abstract

We propose a methodology for hybrid system model reduction that deals with the abstraction of both the continuous and the discrete behaviors of the system. Balanced residualization for continuous dynamics and pseudo-equivalent location elimination for the graph are used for model reduction in the continuous and discrete time domain respectively

Model uncertainty limits the utilization of Model Predictive Controllers (MPC) to minimize building energy consumption. We propose a new Robust Model Predictive Control (RMPC) structure to make a building controller robust to model uncertainty. The results from RMPC are compared with those from a nominal MPC and a common building Rule Based Control (RBC). The results are then used to develop a methodology for selecting a controller type (i.e. RMPC, MPC, and RBC) as a function of building model uncertainty. RMPC is found to be the desirable controller for the cases with an intermediate level (30%-67%) of model uncertainty, while MPC is preferred for the cases with a low level (0-30%) of model uncertainty. A common RBC is found to outperform MPC or RMPC if the model uncertainty goes beyond a certain threshold (e.g. 67%).