Occupancy Detection in a Building Using Hybrid Models

Abstract

Buildings consume over 40% of the world’s total energy supply, and their occupancy is increasingly recognized as a major performance indicator due to its effect on a building’s energy costs and occupant satisfaction. In this paper, a hybrid model is created to estimate future loads of a building with high efficiency and accuracy. The proposed model is composed of two - connected in a cascade - artificial neural networks, where the outcomes of the first network are fed into the second one, which in its turn performs the load forecasts. A pre-existing dataset is used to verify the proposed model and to test a variety of training set sizes. Analysis of the results is executed by taking six pair of combinations separately for both open door and closed door fault cases. In this analysis, cascaded back propagation and Elman back propagation method - among the rest of the analyzed methods – is found to give the best accuracy, i.e, 97.2% - 97.9%, which indicates that the suggested hybrid technique is more accurate than the existing non-hybrid methods.