Abstract
Fluorescence spectroscopy attracted more and more attention in pesticide residue detection field because of its advantages of non-destructive, non-contact, high speed and no requirement of complex pre-process procedure. However, given that the concentration of the pesticide detected via fluorescence spectroscopy is calculated in accordance with the Beer-Lambert law, this method can only be used to detect samples containing a single kind of pesticide or several kinds of pesticides with completely different fluorescence which is not in accordance with practical cases. In this article, to overcome this disadvantage, back-propagation (BP) neural network algorithm was introduced to detect multiple kinds of pesticides via fluorescence spectroscopy. The results from four kinds of pesticides which are usually used for fruits and vegetables indicated the effectiveness of BP neural network algorithm.
Highlights
In order to increase crop production to solve the global starvation problem, varieties of pesticides have been increasing dramatically in recent decades [1], [2]
Given that the concentration of the pesticide detected via fluorescence spectroscopy is calculated in accordance with the Beer-Lambert law, this method can only be used to detect samples containing a single kind of pesticide or several kinds of pesticides with completely different fluorescence which is not in accordance with practical cases
To be different from the training and validation dataset for the model, the train, validation and test divided from the training dataset was called as inner-train, inner-validation and inner-test
Summary
In order to increase crop production to solve the global starvation problem, varieties of pesticides have been increasing dramatically in recent decades [1], [2]. 1) the fabrication of flexible substrates is time-consuming and their shelf life is limited [20]; 2) the instability and worse repeatability of the SERS substrates from different batches which would affect the Raman intensity For these reasons, it seems unsuitable for infrared spectroscopy and Raman spectroscopy to realize on-line detection of the pesticides. Ji et al measured the fluorescence intensity at 356 nm and built the exponential prediction model based on the Beer-Lambert law to predict the concentration of boscalid in grape juice Their results with a correlation coefficient of 0.9999, an average recovery of 101.7%, and a relative standard deviation of 2.5723% indicated that fluorescence spectroscopy is effective to detect the concentration of single pesticide [24]. The results indicated that the predicted concentration fitted well with the actual concentration
Sign-in/Register to view highlights & summary 
Published Version (
Free)
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