Laser-induced breakdown spectroscopy combined with principal component analysis-based support vector machine for rapid classification of coal from different mining areas.

Coal is one of our most important energy sources. Because of differences in coal-forming year and coal seam quality, the elemental compositions of coal vary among different mining areas, directly affecting coal combustion efficiency and pollution emission. To improve the accuracy of coal classification, laser-induced breakdown spectroscopy (LIBS) combined with a support vector machine (SVM) algorithm model based on principal component analysis (PCA) is proposed to identify and classify coal from 10 different regions. The samples were first penetrated using LIBS, and spectral data were collected. The spectral data were then normalized to improve their signal-to-noise ratio. Finally, the PCA-SVM model is established, the model parameters are evaluated using the grid search method, and the spectral feature classification performance is determined via cross-validation. The results show that the classification accuracy of the PCA-SVM model can reach 98.52%. To verify the performance of this classification model, it was compared with those of five other common classification models and proved that PCA-SVM has great potential for coal quality identification and classification, thus providing a new data analysis and processing solution for coal quality identification and classification in industrial applications. [Display omitted] • Efficient coal quality identification and classification method is proposed. • Method is based on laser-induced breakdown spectroscopy combined with SVM and PCA. • Method is experimentally verified using coals from different mining areas. • Results of proposed method are in good agreement with assays from independent lab. • Performance of proposed method is verified against those of other ML-based methods. [ABSTRACT FROM AUTHOR]