Hybrid Model for Building Performance Diagnosis and Optimal Control

Abstract

Modern buildings require continuous performance monitoring, automatic diagnostics and optimal supervisory control. For these applications, simplified dynamic building models are needed to predict the cooling and heating requirement viewing the building as a whole system. This paper proposes a new hybrid model. Half of the model is represented by detailed physical parameters and another half is described by identified parameters. 3R2C thermal network model, which consists of three resistances and two capacitances, is used to simulate building envelope whose parameters are determined in frequency domain using the theoretical frequency characteristics of the envelope. Internal mass is represented by a 2R2C thermal network model, which consists of three resistances and two capacitances. The resistances and capacitances of the 2R2C model are assumed to be constant. A GA (genetic algorithm)-based method is developed for model parameter identification by searching the optimal parameters of 3R2C models of envelopes in frequency domain and that of the 2R2C model of the building internal mass in time domain. As the model is based on the physical characteristics, the hybrid model can be used to predict the cooling and heating energy consumption of buildings accurately in wide range of operation conditions.