Near-infrared spectroscopy, which has advantages of high speed, high precision, and minimal sample preparation, is a rapid, solvent-free, on-site quantitative and qualitative analytical method. This study assessed the feasibility of using near-infrared spectroscopy to determine prochloraz residue in orange juice. Two chemometric methods, partial least-squares discriminant analysis and least-squares support vector machines, were employed to create mathematical models to classify three spectrum bands. The selected spectral values ranged from 5,030 to 5,100 cm−1. The results show the least-squares support vector model outperformed the discriminant analysis algorithm. Classification accuracy of 100% was obtained for the calibration set and the prediction set of the former when the two parameters, γ and σ2 were 20.5707 and 2.48703, respectively, while the classification accuracy of the discriminant analysis method for the calibration set and the prediction set were 100% and 96.8%, respectively. By contrast, it is easy to conclude that when the limit of detection of the proposed method is set as low as 0.5 µg mL−1, the support vector model has the capability to predict the presence or absence of prochloraz residue, whereas the discriminant analysis model may give false positives. Considering the analytical process for a sample takes only about one minute, the proposed method ensures the simplicity and reliability in practical applications. The results of the study showed the potential of near-infrared spectroscopy as a rapid and environmentally friendly method for the on-site screening of prochloraz in orange juice.
Read full abstract