MACHINE LEARNING BASED INDOOR AIR POLLUTANT SOURCE RECOGNITION WITH GAS RESISTANCE AND MULTI-SENSOR ARRAY ELECTRONIC NOSES

Download

10452975 (1).pdf

Date

2022-2-10

Author

Yesilata, Mehmet Yigitcan

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

528
views

352
downloads

Indoor Air Quality (IAQ) is closely linked to health and well-being. Humans spend the majority of their time indoors. Breathable air that is free of harmful pollutants can result in an improved quality of life, a decreased risk of respiratory infections, and a decreased risk of developing chronic conditions. Cleaning chemicals, construction operations, smoking, perfumes, building materials and outdoor pollutants can all contribute to indoor air pollution. Detecting the sources of pollution is essential in order to improve interior air quality. A sensing device known as an electronic nose collects various sensor data to detect scents or flavors. Two different types of electronic nose are used in this research to collect data. The first one has 8 VOC sensors that operate in response to the gas resistance of the MOX layer. Each sensor is simultaneously heated by a different heater profile, and their reaction to the gas produces an 8 dimensional sensitivity that is proportional to the sensors gas resistance at that temperature. The second one collects different types of parameters that affect indoor air quality like PM2.5, PM10, CO2, Formaldehyde, Ethanol, H2, TVOC, Temperature and Humidity with a multi sensor array. The experiment was conducted in a 130-liter box. Throughout the experiment, nine different materials were measured, including office air, smoking, cleaning materials, alcohol-sanitizer, curry, coffee, painted tile, stone wool, and varnished wood with two different electronic noses. Data has been modelled with Naive Bayes, kNN, Random Forest Classifiers to predict pollutant sources in the box. Random Forest algorithm with 10 trees gives the best result with data collected from multi sensor array based electronic nose. The algorithm gives 0.95 accuracy with 0.94 precision on the sensor data.

Subject Keywords

Indoor Air Quality, Electronic Nose, Odor Detection, Machine Learning

URI

https://hdl.handle.net/11511/96757

Collections

Graduate School of Natural and Applied Sciences, Thesis

Suggestions

OpenMETU
Core

Assessment of indoor air quality in crowded educational spaces Betuz, Naima Ebru; Elias Özkan, Soofia Tahira; Department of Building Science in Architecture (2012) Indoor air quality has become a challenge together with the global aim ‘decreasing energy consumption’. Increasing insulation levels of building envelopes but implementing inaccurate building system details has caused excessive heat, accumulation of pollutants, etc. in spaces. In terms of educational spaces, the increase in complaints and illnesses due to unfavorable indoor air conditions leads to a decrease in concentration and so academic performance of students and staff. In the context, the aim of the s...
Assessment of carbon dioxide transmission through porous building materials in relation to indoor air quality Yüncü, Başak; Tavukçuoğlu, Ayşe; Saltık, Emine Nevin; Department of Building Science in Architecture (2016) There is lack of knowledge on breathing features adequacy of porous building materials for sustaining indoor air quality. There is necessity to examine gas diffusion characteristics of breathing porous building materials and the relation of those characteristics with IAQ. A comprehensive study, therefore, was conducted on mud brick as a traditional building material and AAC as a contemporary building material which are well-known by their high breathable characteristics. Their air permeability features were...
Air quality impacts of electricity use in domestic heating: A comparative study for Ankara Kayin, S; Yavuzdemir, F (1997-09-26) In this study, existing air quality affects of current domestic heating systems in comparison to the electricity use alternative in Ankara have been analysed. For the winter season 1991-1992 heat energy used in Ankara for domestic heating has been calculated. For the hypothetical case of electricity use instead of the current domestic heating systems in the city, Industrial Source Complex - Short Term dispersion model has been used for six site alternatives and for two different thermal power plant technolo...
Health impacts of indoor air pollution from household solid fuel on children and women Ali, Muhammad Ubaid; Yu, Yangmei; Yousaf, Balal; Munir, Mehr Ahmed Mujtaba; Ullah, Sami; Zheng, Chunmiao; Kuang, Xingxing; Wong, Ming Hung (2021-08-15) The inefficient and incomplete combustion of solid fuel (SF) is associated with high levels of indoor air pollutants leading to 3.55 million deaths annually. The risk is higher in women and children, due to their higher exposure duration and unique physical properties. The current article aims to provide a critical overview regarding the use of solid fuel, its associated pollutants, their toxicity mechanisms and, most importantly the associated health impacts, especially in women and children. Pollutants as...
Dynamic resistivity behavior of thin oxide based multilayer thin films under reducing conditions Kurbanoğlu, Başak; Karakaş, Gürkan; Department of Chemical Engineering (2006) Effects of urban air pollution on health and environment have lead researchers to find economic air quality monitoring regulations. Since tin dioxide (SnO2) was demonstrated as a gas sensing device in 1962, tin oxide based thin film sensors have been widely studied due to their high sensitivity and fast response. The main advantages of using tin oxide sensors are their low cost, small size and low power consumption for mobile system applications. But, in order SnO2 based sensors to meet low concentration of...

Citation Formats

M. Y. Yesilata, “MACHINE LEARNING BASED INDOOR AIR POLLUTANT SOURCE RECOGNITION WITH GAS RESISTANCE AND MULTI-SENSOR ARRAY ELECTRONIC NOSES,” M.S. - Master of Science, Middle East Technical University, 2022.