A novel Local Correction-based (LC-b) approach for low-order control-oriented models

Abstract

The development of a model is a crucial step for Model-based Predictive Control (MPC) strategies. Modelling is also the phase involving the most time and effort. Low-order models represent a practical solution to reduce time, information and technical expertise required to develop a control-oriented approach. However, it is essential to guarantee a reasonable prediction accuracy while keeping the model simple. In this paper, a novel Local Correction-based (LC-b) approach, which improves the prediction accuracy of low-order models of thermal zones, is presented. The LC-b approach enables accounting for local thermal effects, thus improving the predictions of the indoor air temperature and heating load. The proposed approach was used to model a two-storey building using operation measurements; the accuracy of predictions was evaluated in terms of statistical indices. Results show that the proposed LC-b approach improved the predictions from a conventional low-order model by up to 68%.