How do different locations, floors and aspects influence indoor radon concentrations? An empirical study using neural networks for a university campus in Northwestern Turkey
Yükleniyor...
Dosyalar
Tarih
2013
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Sage Publications Ltd
Erişim Hakkı
info:eu-repo/semantics/closedAccess
Özet
Indoor radon (Rn-222) concentrations were measured at a 10-min interval during October 2011 and January 2012. The monitoring followed a randomised and repeated pattern of experimental design, and was carried out at six faculty buildings of the Abant Izzet Baysal University, on five floor levels and two aspect directions (south vs. north) using an AlphaGUARD P30 Radon Monitor. The University campus area located in northwestern Turkey is near the North Anatolian Fault, a major active right lateral-moving strike-slip fault which runs along the transform boundary between the Eurasian Plate and the Anatolian Plate. Best artificial neural networks (ANNs) emulating indoor Rn-222 levels were selected as a function of air temperature (T-a), air pressure (P-a), relative humidity (RH), T-a by RH interaction, local time, location, floor and aspect. Elevated levels of indoor Rn-222 concentrations were measured at the south-facing offices and on the first floor levels of the building. Lower concentrations were found on the upper floor levels. Out of 27 ANNs, GFF-1-B-L and MLP-2-B-L performed best and could be contributing to the 35.6% and 87.2% of variations in spatio-temporal dynamics of indoor Rn-222 levels as a function of location or floor level and aspect, respectively, in addition to T-a, P-a, RH, T-a by RH interaction and time.
Açıklama
Anahtar Kelimeler
Indoor Radon, Abant Izzet Baysal University, Neural Networks, Spatio-temporal Dynamics, Modelling
Kaynak
Indoor And Built Environment
WoS Q Değeri
Q1
Scopus Q Değeri
Q2
Cilt
22
Sayı
4
