Simplified building model for transient thermal performance estimation using GA-based parameter identification

Building simple and effective models are essential to many applications, such as building performance diagnosis and optimal control. Detailed physical models are time consuming and often not cost-effective. Black box models require large amount of training data and may not always reflect the physical behaviors. In this study, a method is proposed to simplify the building thermal model and to identify the parameters of the simplified model. For building envelopes, the model parameters can be determined using the easily available physical details based on the frequency characteristic analysis. For the building internal mass involving various components, it is very difficult to obtain the detailed physical properties. To overcome this problem, the building internal mass is represented by a thermal network of lumped thermal mass and the parameters are identified using operation data. Genetic algorithm (GA) estimators are developed to identify these parameters. The simplified dynamic building energy model is validated on a real commercial office building in different weather conditions.