Nondestructive and rapid grading of tobacco leaves by use of a hand-held near-infrared spectrometer, based on a particle swarm optimization-extreme learning machine algorithm

Abstract

A nondestructive and rapid method has been put forward to grade tobacco leaves in the paper. The method is based on a combination of a hand-held near-infrared spectrometer and a particle swarm optimization-extreme learning machine algorithm. Firstly, the spectral data of the training samples are collected directly from the tobacco leaves nondestructively by using a hand-held near infrared spectrometer without any pretreatment. Secondly, the training models of different classes are built using particle swarm optimization-extreme learning machine algorithm. Finally, the classes of test samples can be predicted by using the developed models. Besides, the classification results of particle swarm optimization-extreme learning machine algorithm are also compared with that of the traditional linear discriminant analysis, support vector machine, and extreme learning machine algorithms, respectively. The experimental results show the classification accuracy of the particle swarm optimization-extreme learning machine algorithm is comparable after the parameter optimization. It indicates that the interplay between the hand-held near-infrared spectroscopy technology and particle swarm optimization-extreme learning machine algorithm will provide a novel classification method for grading tobacco leaves in the purchasing process on the spot.