Paper Conference

Proceedings of Building Simulation 2021: 17th Conference of IBPSA

     

Effect of limited controller and sensor datasets on the performance of data-driven model predictive control for residential buildings

Hyeong Seok Lee, Yeonsook Heo
Urban Energy&Environment Lab, Dept. of Civil, Environmental and Architectural Engineering, Korea University

DOI: https://doi.org/10.26868/25222708.2021.30750
Abstract: Data-driven model predictive control (MPC) has gained popularity as a proactive control strategy due to the development of the Internet of Things (IoT) and data storage technologies. To apply the data-driven MPC to residential buildings, it is necessary to consider what data to measure, how to control a target system, and which prediction models are inserted. In this paper, we investigate the effect of different datasets due to sensor and controller set selection on data-driven model and data-driven MPC performance. We confirmed that the control variable according to the controller set affects MPC performance considerably.
Keywords: Data-driven model predictive control, Autoregressive with exogenous input model, Linear Programming, Residential buildings
Pages: 2671 - 2678
Paper:
bs2021_30750