Stand-off laser-induced breakdown spectroscopy (ST-LIBS) has developed into an excellent technology for remote analysis of geological samples. In this study, the rock classification capability of principal component analysis (PCA) was illustrated. The method of automatic spectral peaks identification with linear discriminant analysis (ASPI-LDA) was first applied for classification of rocks and compared with manual spectral peaks identification with linear discriminant analysis (MSPI-LDA). The spectra of rocks were obtained using Echelle and Czerny-Turner spectrometers at 5 m distance and the spectrum peaks were automatically and rapidly identified with ASPI-LDA, not manually identified one by one. The results suggested that MSPI-LDA outperformed PCA in classification performance and, the predictive classification accuracies were further significantly improved by ASPI-LDA with two spectrometers. Besides, the performances of two spectrometers were compared. The results showed that the ASPI-LDA algorithm remedied for the weakness of a compact Czerny-Turner spectrometer having a narrow spectral range to achieve high accuracy classification. Meanwhile, the compact spectrometer used for field detection in ST-LIBS was miniaturized and of lower cost. This indicates that ASPI-LDA combined with a compact spectrometer has a great potential for field in situ remote detection in ST-LIBS.
|Original language||English (US)|
|Number of pages||7|
|Journal||Journal of Analytical Atomic Spectrometry|
|State||Published - Mar 2018|
ASJC Scopus subject areas
- Analytical Chemistry