Rapid origin identification of chrysanthemum morifolium using laser-induced breakdown spectroscopy and chemometrics

Abstract

Chrysanthemum is widely grown throughout China, but there are large differences in quality, chemical composition and price among chrysanthemum from different origins, and similar appearance traits make it difficult to distinguish chrysanthemum. In this study, laser-induced breakdown spectroscopy (LIBS) technology combined with chemometrics was successfully used to classify chrysanthemum from nine different geographical sources. Different data processing methods and feature variable selection methods were evaluated, and the first derivative combined with the least absolute shrinkage and selection operator algorithm (1st Der-LASSO) was finally selected as the best processing method. The good classification performance was obtained using linear discriminant analysis (LDA), K-nearest neighbors (KNN), multi-layer perceptron (MLP) and random forest (RF) models with prediction accuracies of 100.0%, 99.0%, 96.5%, 99.4% for the validation set, respectively. For the independent test set, the established LASSO-LDA model can achieve 85.9% prediction accuracy. Furthermore, according to the importance of variables calculated by the RF model, Fe was the most important element to distinguish chrysanthemum from different origins, and the order of importance of other elements was O, Ca, Cu, K, Cl, B, Mg, Na. The combination of LIBS and chemometrics provides a simple, fast and reliable method for the geographic origin classification of chrysanthemum samples. This study provides a new basis for the application of LIBS technology in origin identification in the food field.