ABSTRACT
ABSTRACT: Aflatoxins content in sweet corn at early stages was quantitatively detected by an improved electronic nose. After storing 96 h at 30°C and 75% RH, the aflatoxins concentration measured by the competitive direct enzyme-linked immunosorbent assay (CD-ELISA) was 4.32 ppb that was over the legal limit (4 ppb). The loading analysis and Cluster Analysis (CA) were applied to optimize sensor arrays, and the Multilayer Perceptron Neural Network (MLPNN) was also applied to aflatoxins content. After optimizing the sensor arrays, the correlation coefficient of the prediction and real aflatoxins content increased from 0.99636 to 0.99871, and the average relative error decreased from 1.58% to 0.92%. The excellent performance of the improved electronic nose makes it a good choice to play as a rapid, easy operation, low cost and effective method to quantify aflatoxins content in sweet corn at early stages.
