Nonlinear Predictive Control of Chiller System using Gaussian Process Model

For Nonlinear Model Predictive Control (NMPC) to be implemented in real application, data driven models are advantageous since they can be easily constructed and are relatively fast, compared to first principle based models (simplified calculation [ISO 13790], dynamic simulation [EnergyPlus, ESP-r, TRNSYS, etc.], state space models, etc.). Gaussian Process Model (GPM), one of the data-driven approaches, can be beneficially used for real time stochastic optimal control of nonlinear building systems, since the GPM is very lightweight in terms of computation time and does not require significant modeling efforts. The GPM is a black-box model based on Bayesian approach. For real-time optimal control of chiller operation in an office building, the authors developed a coupling between the GPM and an optimization routine (Genetic Algorithm) in MATLAB optimization toolbox. The two control parameters are studied in the paper: outlet temperatures of a chilled water loop as well as a cooling tower loop respectively. This study delivers real-time optimal outlet temperatures of the chilled water loop and cooling tower loop. In addition, the characteristics of GPM for reliable NMPC were discussed in the paper. It was shown that GPM produces satisfactory control performance taking into account the probabilistic nature of the chiller system.