Introduction of a plug and play model predictive control to predict room temperatures. (November 2021)

• Record Type: Journal Article
• Authors: Junghans, Lars; Woo, Deok-Oh
• Is Part Of: Journal of building engineering. Volume 43(2021)
• Publication Date: 2021-11
• Subjects: Room temperature prediction -- Model predictive control -- Plug and play -- Climate surface model; Building -- Periodicals; 690.05
• DOI: 10.1016/j.jobe.2021.102578
• Languages: English
• ISSNs: 2352-7102

Abstract

Abstract: Model Predictive Control (MPC) framework that predicts future room conditions have been examined in the past. However, the technique in its current state has been found to have entry barriers for its use in real building environments because the implementation of physical room models into MPC control framework is time-consuming. Additionally, cost intensive expert knowledge is needed for its implementation. In this study, a "Plug and Play" MPC framework for building automation environments is introduced that predicts the room temperature, energy demand values and load values for the next day. The novelty of the MPC framework is its ability to calibrate to the physical environment in relatively short time periods (10 days) without providing previous information. A calibrated simulation model approach is combined with a rule-controlled filter and is integrated into the MPC framework. Results illustrate that the framework predicts a reliable range of data for the next day. Highlights: Introduction of a "Plug and Play" Model Predictive Control (MPC) framework for a single room. Model Predictive Control (MPC) framework that predicts the future room temperature and energy demand by using the calibrated simulation model approach. The novelty of the MPC framework is its ability to calibrate to the physical environment in relatively short time periods (10 days) without providing previous information. The proposed MPC framework is able to overcome the entry barriers of the MPC technology for room temperature predictions. The accuracy of the MPC framework have been evaluated by using the mean squared error MSE method for varying time periods and seasons.