Model reduction and nonlinear MPC for energy management in buildings

Abstract

In this paper, we introduce a singular perturbations framework for the dynamic analysis and model reduction of building models. Working with a prototype building, we present a theoretical justification of the empirically acknowledged multiple time scale dynamic response of buildings, and develop a mathematically rigorous methodology for deriving reduced-order models for the dynamics in each time scale. Our analysis accounts for the potential use of Heat Recovery Ventilators (HRVs), and we show that their presence leads to the emergence of a dynamic behavior with three time scales, including an overall, system-wide component which involves both the building and the HVAC system. The second part of the paper presents a simulation case study, where we demonstrate the use of the derived reduced-order models in the synthesis of a nonlinear predictive model-based optimal energy management strategy for a single-zone test building situated on the University of Texas campus. The proposed controller exhibits excellent performance, can easily be executed in realtime and its application results in significant energy savings compared to setpoint tracking strategies.

Regression Models for Output Prediction of Thermal Dynamics in Buildings

“…The increased demand for electricity power and/or fuel consumption in residential buildings requires an efficient control design for all heating/cooling equipment. To this end, recently, there have been several efforts towards a more efficient climate control in residential buildings [2,3,4,5]. Ef- * This work is part of the mpcEnergy project which is supported within the program Regionale Wettbewerbsfähigkeit OÖ 2007-2013 by the European Fund for Regional Development as well as the State of Upper Austria.…”

“…In reference [12], a comparison is performed between standard linear ARX models with two time-scale transfer models, which according to the authors better represent thermal models. Indeed, the operation of either an radiant-heating (RH) system or an HVAC system usually takes place in a faster time-scale than the room dynamics, an observation that was also utilized for the derivation of a (nonlinear) model predictive controller in [5] using singularperturbation techniques.…”

Generalized online transfer learning for climate control in residential buildings

“…This observation has been pointed out by several authors, leading to more detailed identification schemes, such as the multiple-time scale analysis presented in [4], the more detailed models of HVAC systems discussed in [10] and the nonlinear regression models developed in [5].…”

“…Recent studies on climate control (heating/cooling) in residential buildings have demonstrated the importance of accurate system identification and prediction for energy savings [2,3,4,5,6]. In fact, there have been several efforts on exploiting the benefits of such prediction schemes through the development of model-predictive-control (MPC) approaches [2,3,4], where predictions of the temperature evolution, weather conditions and user behavior can be incorporated directly into the control design.…”

“…In fact, there have been several efforts on exploiting the benefits of such prediction schemes through the development of model-predictive-control (MPC) approaches [2,3,4], where predictions of the temperature evolution, weather conditions and user behavior can be incorporated directly into the control design.…”