UMP Institutional Repository

Indoor occupancy estimation using carbon dioxide concentration and neural network with random weights

Muhammad Faris, Ramli and Kishendran, Muniandy and Asrul, Adam and Ahmad Fakhri, Ab. Nasir and Mohd Ibrahim, Shapiai (2020) Indoor occupancy estimation using carbon dioxide concentration and neural network with random weights. In: IOP Conference Series: Materials Science and Engineering, The 6th International Conference on Software Engineering & Computer Systems, 25-27 September 2019 , Pahang, Malaysia. pp. 1-9., 769 (012011). ISSN 1757-899X

[img]
Preview
Pdf
Indoor occupancy estimation using carbon dioxide.pdf
Available under License Creative Commons Attribution.

Download (554kB) | Preview

Abstract

This study presents the indoor occupancy estimation using carbon dioxide concentration and neural network with random weights (NNRW). The utilization of carbon dioxide concentration is as an alternative to overcome the limitation of existing techniques, such as dependency to favourable lighting condition and camera position. Whereas, NNRW provides a generalized and fast learning speed classification. In this study, MH-Z19 sensor is used to acquire carbon dioxide concentration and the NNRW is a multiclass estimation method. The numbers of the occupants are divided into three different classes, which are 15 occupants, 30 occupant and 50 occupant classes. Result indicates that the NNRW classifier has obtained training and testing accuracy, about 100 percent and 52 percent, respectively.

Item Type: Conference or Workshop Item (Lecture)
Uncontrolled Keywords: Neural Network With Random Weights (NNRW); MH-Z19; Camera Position
Subjects: Q Science > QA Mathematics > QA76 Computer software
Faculty/Division: Institute of Postgraduate Studies
College of Engineering
Faculty of Manufacturing and Mechatronic Engineering Technology
Depositing User: Pn. Hazlinda Abd Rahman
Date Deposited: 23 Dec 2020 02:53
Last Modified: 23 Dec 2020 02:53
URI: http://umpir.ump.edu.my/id/eprint/27768
Download Statistic: View Download Statistics

Actions (login required)

View Item View Item