Odor measurement and intelligent classification

A new electronic nose system is proposed based on a neural network. The neural network used here is a competitive neural network using the learning vector quantization. Various odors are measured by an array of many metal-oxide semiconductor gas sensors. The noises are reduced by preprocessing the odor data which are measured under the different concentration. Maximum values among the time series data of odors are used. Since the data are affected by concentration levels, a normalization method to reduce the fluctuation of the data is applied. Those data are used to classify the various odors of teas and coffees. The classification accuracy is around 96% in case of four kinds of teas and around 89% for five kinds of coffees.