Cereal Chem. 73 (4):457-461 |
Analytical Techniques and Instrumentation
Electronic Nose for Odor Classification of Grains.
T. Börjesson (1,4), T. Eklöv (2), A. Jonsson (1), H. Sundgren (2), and J. Schnürer (3). (1) Swedish Farmers Supply & Marketing Association, Box 30192, S-104 25 Stockholm, Sweden. (2) Laboratory of Applied Physics, Linköping University, S-581 83 Linköping, Sweden. (3) Department of Microbiology, Swedish University of Agricultural Sciences, S-750 07 Uppsala, Sweden. (4) Corresponding author. Fax: 46 510 664 38. Accepted April 16, 1996. Copyright 1996 by the American Association of Cereal Chemists, Inc.
An electronic nose was used to classify grain samples based on their smell and to predict the degree of moldy/musty odor. A total of 235 samples of wheat, barley and oats, which had been odor classified by at least two grain inspectors, were used. Headspace samples from heated grain were pumped through chambers containing metal oxide semiconductor field effect transistor (MOSFET) sensors, SnO(2) semiconductors and an infrared detector monitoring CO(2). The sensor signals were evaluated with a pattern-recognition software program based on artificial neural networks. The samples were divided into either the four classes moldy/musty, acid/sour, burnt, or normal or the two classes good and bad according to the inspectors descriptions. They were also assigned a score describing their intensity of moldy/musty odor. The electronic nose correctly classified approximately 75% of the samples when using the four-class system and approximately 90% when using the two-class system. These values exceeded the corresponding percentages of agreement between two grain inspectors classifying the grain.