Non-Linear and Linear Baseline Energy Modelling Comparative Studies in Educational Buildings

Baseline energy model is a powerful tool in describing energy consumption for any buildings at a certain period of time. Baseline energy model main task is to predict the energy consumption if energy conservation measures were implemented for energy saving calculation purposes. A simple and practical approach to model a baseline energy utilizing linear regression method is undeniable to be inferior. Linear regression work best with energy consumption that behaves linearly but a practical and simple approach to model a baseline energy for energy consumption that behaves in a non-linear manner have to be establish. Thus, this paper main intention is to model a baseline energy using Non-Linear Auto Regressive with Exogenous Input Model with Artificial Neural Network (NARX-ANN) method as the model estimator. Two buildings in a University compound will be used to model the baseline energy consumption. Multiple regression model will be used in order to compare the results of the baseline energy models developed with NARX-ANN method. It is found that NARX-ANN performs better in terms of error measurement. It is hoped that a better baseline energy models can be developed in order to provide less error if energy saving calculation purpose will be implemented.