Model Identification for the Control of Naturally Ventilated Buildings

In this paper, predictive models are developed to enable the application of model predictive control (MPC) to naturally ventilated buildings. The essential component of an MPC strategy is the predictive model of the building’s thermal dynamics, which is the focus of this study. An empirical approach is taken using multilayer perceptron (MLP) neural network models. The models presented were generated using data gathered from real buildings during operation and building simulation data generated using EnergyPlus. The resulting models were able to accurately predict internal conditions such as zone temperature. The problem of insufficient input excitation is highlighted and an identification procedure to overcome it is presented.