Chemometrics; Classification; Laser-induced breakdown spectroscopy; Variety; Walnut powder;JUGLANS-REGIA L.; SUPPORT VECTOR MACHINE; CLASSIFICATION; LIBS; CHEMOMETRICS; VARIABILITY; EMISSION; SAMPLE; OILS
Rapid classification of varieties of walnut powder is important to protect consumers against potential fraud because different varieties of walnut powder vary dramatically in quality, taste, and price. Laser-induced breakdown spectroscopy (LIBS) is a promising technology for on-line application that has the advantages of no or little sample preparation, fast response (usually seconds to minutes), and remote detection capability. In this research, LIBS and multi-variable analysis methods were applied to classify varieties of walnut powder for the first time. Before LIBS analysis, a fairly flat surface was provided to reduce fluctuation, and no other sample preparation was needed. After acquiring the LIBS spectra, principal component analysis (PCA) was used to make qualitative discrimination, and classification models were developed with partial least squares discrimination analysis (PLS-DA) and support vector machine (SVM). It was shown that three origins as well as different varieties from the same origin could be differentiated successfully by LIBS and chemometrics, while the discrimination of six different varieties with a single ""global"" model was not very successful. SVM could help to handle complex LIBS dataset, which improved the classified rate. The presented results provide the first proof-of-principle data for on-line application of the LIBS technique for classification of varieties of walnut powder.