Abstract
Microwave detection technology is innovatively applied to the rapid detection of the aflatoxin B1 (AFB1) in wheat. A miniaturized microwave detection device based on free-space measurement was developed and used to acquire the transmission index in the frequency band of 2.5 GHz to 11.5 GHz for wheat samples with different mildew levels. Bootstrapping soft shrinkage (BOSS) was applied to optimize the transmission index after the least square filtering. The linear model (partial least squares, PLS) and non-linear models (i.e., support vector machine, SVM; extreme learning machine, ELM; random forest, RF) based on the optimized feature variables were constructed separately for quantitative measurement of the AFB1 concentration in wheat. The results showed that the BOSS can generate highly targeted feature variables. Moreover, the BOSS-SVM model achieved the optimal prediction performance. In the prediction set, the root mean square error of prediction (RMSEP), the coefficient of determination (RP2) and the relative prediction deviation (RPD) of the model were 2.8 μg·kg−1, 0.97, and 5.7, respectively. The study demonstrates that the monitoring of the AFB1 concentration in wheat can be achieved with high accuracy using the self-made microwave detection device in combination with an appropriate chemometric method. This study may provide a new technical approach for on-site monitoring of quality and safety of stored grains.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call