Share

Practice-oriented system identification strategies for MPC of building thermal and HVAC dynamics.

Number: pap. 3142

Author(s) : ALANQAR A., ELLIS M. J., TAPIERO J. E.

Summary

In this paper, we present two practice-oriented strategies to overcome difficulties in conducting system identification of building thermal and HVAC dynamics. Specifically, the first part of the paper presents a saturation detection and removal algorithm for buildings HVAC systems. This work focuses on identifying linear time invariant (LTI) models of building thermal and HVAC dynamics, and such models are not able to capture the nonlinear dynamics associated with controller saturation. For simplicity, data collected during periods where the controller was saturated is referred to as saturation data. The purpose of this algorithm is to detect and remove saturation data that are not suitable for system identification, and thereby utilize only the reliable excited data that produce more accurate models of buildings HVAC systems. The second part of this paper presents a systematic procedure that eliminates poor initial guesses when solving the non-convex system identification problem to identify building thermal and HVAC model parameters. The procedure presents guidelines to tests the quality of initial guesses and yields guesses that ultimately lead to models that accurately fit the data and captures the system dynamics. Although these strategies can be implemented to both black-box and grey-box system identification, we will present them in the context of grey-box. The proposed strategies were implemented to on a real building in cooling mode and resulted in very accurate models.

Available documents

Format PDF

Pages: 10

Available

  • Public price

    20 €

  • Member price*

    15 €

* Best rate depending on membership category (see the detailed benefits of individual and corporate memberships).

Details

  • Original title: Practice-oriented system identification strategies for MPC of building thermal and HVAC dynamics.
  • Record ID : 30024736
  • Languages: English
  • Subject: Environment
  • Source: 2018 Purdue Conferences. 5th International High Performance Buildings Conference at Purdue.
  • Publication date: 2018/07/09

Links


See other articles from the proceedings (88)
See the conference proceedings

Indexing