A State-Space Modeling Approach and Subspace Identification Method for Predictive Control of Multi-Zone Buildings with Mixed-Mode Cooling

Key

3636

Conference Year

2014

Authors

Jianjun Hu, School of Civil Engineering and Division of Construction Engineering and Management, Purdue University, United States of AmericaFollow
Panagiota Karava, School of Civil Engineering and Division of Construction Engineering and Management, Purdue University, United States of AmericaFollow

Keywords

Model Predictive Control, Mixed-mode Cooling, Prediction Model, Linear Time-Variant State-Space Model, Model Identification

Abstract

The paper presents a control-oriented modeling approach for multi-zone buildings with mixed-mode (MM) cooling that incorporates their mode switching behavior. A forward state-space representation with time-varying system matrices is presented and used for establishing a detailed prediction model of a multi-zone MM building. The linear time-variant state-space (LTV-SS) model, which is considered as a true representation of the building, is used for developing data-driven linear time-invariant state-space models based on the subspace identification algorithm. The simplified black-box model can successfully capture the switching behavior of the MM building with the RMSE of 0.64 ºC.