A Simplified Model Parameter Identification Methodology for Buildings Indoor Thermal Behavior Control: A Case Study in a Tropical Climate of Panama

Authors

Rivera, Ana

Sánchez, Josue

Chen Austin, Miguel

Abstract

The design of building control strategies that maintain occupant comfort and improve energy efficiency relies on a thermal model to predict indoor temperature. The model should represent the building envelope and its systems. Other than controller design, the thermal model is useful for the evaluation and simulation of energy optimization strategies. In this paper a methodology to identify building thermal models, based on a thermal-electrical analogy, is proposed. The methodology can be applied to residential case studies, located in tropical climates. It consists of determining which candidate model best describes the building thermal dynamics. The models are trained multiple times to study parameter estimate dispersion, and if the estimates converge to a single value regardless of their initial value, the models are validated. The model with the lowest root-mean-square error (RMSE) is selected as the best model. When the methodology is tested in a residential case study located in Panama, the best network has a validation RMSE of 0.36°C, which is satisfactory for controller design purposes. The model is then used to tune a proportional integral derivative (PID) controller which is then successfully employed to maintain a desired indoor temperature. Tuning the controller with the identified model avoids the need for tedious trial and error controller tuning.