A Numerical Approach for Buildings Reduced Thermal Model Parameters Evaluation

Abstract

This paper introduces an efficient PSO-based optimization method to identify parametric values of buildings reduced thermal RC network models. A high order reference model representing energy flow through building envelope is developed based on Crank-Nicolson finite difference method. Thermal network models performance with random resistance and capacitance values are compared with a reference model then PSO is used for parameters identification to match the actual reference model thermal dynamics. The accuracy and computational cost of these optimized models are validated against reference model for different construction classes. The proposed methodology presents an effective way to develop a reliable thermal network model representing realistic thermal dynamics in building that can be used to improve efficiency of controlling techniques.