Coal quality regression data by machine learning and LIBS

Laser-induced breakdown spectroscopy (LIBS) was utilized to simultaneously determine the elemental carbon content, ash content, volatile matter, total sulfur, and gross calorific value of coal samples by establishing quantitative algorithms between LIBS spectra and values tested by standard methods of 49 coal samples. The quantitative analysis performance of support Vector Machine (SVM), Random Forest (RF), Kernel Extreme Learning Machine (K-ELM), and Least Squares Support Vector Machine (LS-SVM) after preprocessing of abnormal data removal, baseline correction and noise reduction.