A comparative model combining carbon atomic and molecular emissions based on partial least squares and support vector regression correction for carbon analysis in coal using LIBS

The aim of this study was to analyze the carbon contents in coal samples by laser-induced breakdown spectroscopy (LIBS). The 266 nm laser radiation was utilized for laser ablation and plasma generation under atmospheric conditions. The correlated carbon atomic and molecular emission lines were determined for the variables of the multiple linear regression (MLR) model. Then, the plasma temperatures of different coal samples were compared to characterize the necessity of residue correction from the MLR model. Finally, the partial least squares regression (PLSR) and support vector regression (SVR) were proposed to correct the residue errors of the MLR model. R 2 , RMSECV, and RMSEP for the MLR model were 0.86%, 3.20%, and 3.41%, whereas these values for the MLR model coupled with the PLSR correction model were 0.99%, 0.13%, and 2.46%, respectively; moreover, these values for the MLR model coupled with the SVR correction model were 0.99%, 0.00039%, and 1.43%. The results showed that the combination of carbon atomic and molecular emissions with both PLSR and SVR correction could improve the measurement accuracy, and the SVR correction helped in achieving better measurement accuracy. This paper proposed an innovation model combining carbon atomic and molecular emissions based on support vector regression correction for quantitative analysis of carbon in coal using LIBS.