Graphy Publications
Inspiring Innovations & Discoveries
International Journal of Mechanical Systems Engineering Volume 1 (2015), Article ID 1:IJMSE-106, 9 pages
http://dx.doi.org/10.15344/2455-7412/2015/106
Research Article
Occupancy Estimation in a Subway Station Using Bayesian Simulation Based on Carbon Dioxide and Particle Concentrations

Cheolyong Shin1 and Hwataik Han2*

1Graduate School, Kookmin University, 77 Jeungneung-ro, Seongbuk-gu, Seoul 136-702, Korea
2Department of Mechanical Engineering, Kookmin University, 77 Jeungneung-ro, Seongbuk-gu, Seoul 136-702, Korea
Dr. Hwataik Han, Department of Mechanical Engineering, Kookmin University, 77 Jeungneung-ro, Seongbuk-gu, Seoul 136- 702, Korea, Tel: 82-10-7211-4687; E-mail: hhan@kookmin.ac.kr
01 July 2015; 31 October 2015; 01 November 2015
Shin C, Han H (2015) Occupancy Estimation in a Subway Station Using Bayesian Simulation Based on Carbon Dioxide and Particle Concentrations. Int J Mech Syst Eng 1: 106. doi: http://dx.doi.org/10.15344/2455-7412/2015/106
This work was supported by the BK-Plus21 Program (31Z20130012959) of the Korea Research Foundation, and by the Human Resources Development Program (20134040200580) of the Korea Institute of Energy Technology Evaluation and Planning (KETEP), funded by the Ministry of Trade, Industry and Energy, Republic of Korea.

Abstract

Demand controlled ventilation is an energy saving approach used to regulate outdoor air supply to a space according to its demand. The occupancy within a space is a useful parameter reflecting the ventilation requirement. The objective of the present study is to develop a method for estimating the occupancy in a subway station based on CO2 and PM10 concentration data using the Indoor Air Quality Tele-Monitoring System located in the station. A feasibility study has been conducted to investigate the monitoring system can provide occupancy information with satisfactory accuracy for ventilation control purposes. Bayesian inference is used in estimating the occupancy at a platform based on unknown information such as ventilation rate and CO2 generation rate per person using various assumptions. The posterior distribution of the occupancy was simulated using the Markov Chain Monte Carlo sampling method. The results indicate that the dynamic model reduces the effect of the time delay and improves the uncertainty bands in the occupancy inference more than the static model. The inferred occupancy results are within the uncertainty ranges of the actual occupancy in the station. Additional use of the PM10 concentration data improves the accuracy of the inference further.