A data mining model for building occupancy estimation based on deep learning methods. Issue 7 (September 2019)

• Record Type: Journal Article
• Authors: Zhou, Yaping; Yu, Zhun (Jerry); Li, Jun; Huang, Yuanjian; Zhang, Guoqiang
• Is Part Of: IOP conference series. Volume 609:Issue 7(2019)
• Publication Date: 2019-09
• Subjects: Materials science -- Periodicals; 620.1105
• DOI: 10.1088/1757-899X/609/7/072029
• Languages: English
• ISSNs: 1757-8981

Abstract

Abstract: Real-time occupancy estimation is of great importance to improve systems control and energy efficiency of buildings. This study proposed a novel real time occupancy estimation model based on CO2 concentration data. A non-neural-network deep learning method (i.e. gcForest) was used to estimate the number of occupants. The gcForest incorporates three classifiers in each level, enabling the estimation performance to be enhanced by exploiting the complementarity among different learning algorithms. To evaluate the effectiveness of the proposed model, this study conducted an experiment in an office room and compared its results with the IHMM model that was widely used in previous studies. The experimental results indicate that the proposed model could achieve higher estimation accuracy and higher detection accuracy of occupant presence or absence.