Intelligent Electronic Nose System Independent on Odor Concentration

Abstract

Conventional odor classification methods have been considered under steady state conditions of temperature, humidity, density, etc. In real applications, those conditions may not occur and they will be variable from time to time. Therefore, it is necessary to find some features which are independent on various environmental conditions. In this paper, using a derivative of odor density and odor sensing data in log scale, we will find a feature of the odor which is independent on the density. Using this feature, we construct an electronic nose which is independent on odor density based on a neural network. The neural network used here is a competitive neural network by the learning vector quantization (LVQ). Various odors are measured with an array of many metal oxide gas sensors. After removing noises from the odor data which are measured under the various concentrations, we take the maximum values among the time series data of odors. to reduce the effect of concentration, we use a normalization method to reduce the fluctuation of the data due to the concentration levels. Those data are used to classify the various odors of tees and coffees. The classification results show the effectiveness of the proposed method.