Quantitative analysis of different volatile organic compounds using an improved electronic nose

Abstract

This paper sets up an improved electronic nose with an automatic sampling mode, large volumetric vapors and constant temperature for headspace vapors and gas sensor array. In order to facilitate the fast recovery and good repeatability of gas sensors, the steps taken include (A) short-time contact with odors measured; (B) long-time purification using environmental air; (C) exact calibration using clean air before sampling. We employ multiple single-output perceptrons to discriminate and quantify multiple kinds of odors. This task is first regarded as multiple two-class discrimination problems and then multiple quantification problems, and accomplished by multiple single-output perceptrons followed by multiple single-output perceptrons. The experimental results for measuring and quantifying 12 kinds of volatile organic compounds with changing concentrations show that the type of electronic nose with a hierarchical perceptron model has a simple structure, easy operation, good repeatability and good discrimination and quantification performance.