Towards the development of an electronic nose for general odor classification

• Main Author: ROA, MARY ANNE
• Format: text
• Published: Archīum Ateneo 2017
• Subjects: Olfactometry; Olfactory sensors; Odors; Biosensors; Neural networks (Computer science); Detectors > Design and construction
• Online Access: https://archium.ateneo.edu/theses-dissertations/84; http://rizalls.lib.admu.edu.ph/#section=resource&resourceid=1397784320&currentIndex=0&view=fullDetailsDetailsTab
• Institution: Ateneo De Manila University

The electronic nose is an emerging field of electronics, mechatronics, robotics and a promising application of artificial intelligence. It has received considerable attention in the field of sensor technology during the past two decades. Some of its potential uses include identification of toxic wastes, quality control, monitoring air quality, examining odors in infected wounds and inspection of food. Notwithstanding the vast amount of literature on the usage of electronic noses for specific purposes, this technology originally and ultimately aims to mimic the capability of mammals to discriminate odors from all sorts of objects. This study demonstrated the theoretical and practical feasibility of designing an electronic nose towards general odor classification. A multi-sensor array hardware unit was carefully constructed in this study for data collection. Important hardware design considerations such as sensor calibration, aeration, circuit protection and voltage/current requirements were satisfied. A highly fine-tuned artificial neural network (ANN) was integrated to this hardware to interpret and relate the data to a target odor class. Various network architecture considerations, such as neuron count, number of layers and activation function, as well as various data treatment methods, such as normalization, division and cross-validation, were investigated before coming up with the final neural network. The result showed that careful hardware integration with an ANN having sufficiently deep internal structure is able to return accurate prediction to at least half of the ten general odor classes proposed by the researchers from University of Pittsburg and Bates College [14] namely fragrant (96%), fruity (98%), chemical (98%). peppermint (98%) and popcorn (90%). This demonstrates the feasibility of making electronic noses for general odor classification which could lead to further broadening of e-nose applications.