Determination of Calorific Value of Mixed Coals by Analysis of Major Elements Using Data Pre-Processing in Laser-Induced Breakdown Spectroscopy

The purpose of this study was to enhance the accuracy of the calorific value estimation of coal by applying data preprocessing methods in laser-induced breakdown spectroscopy (LIBS). The Savitzky–Golay (SG)-smoothing and SG derivative preprocessing methods were adopted to improve the accuracy of the prediction model. The relationship among the original, SG-smoothing-pretreated, and SG derivative-pretreated LIBS data and their elemental concentrations were determined using the partial least squares regression (PLSR) model. In order to compare the reliability of each PLSR model, the coefficient of determination, root mean square error (RMSE), relative error, and RMSE average were used. As a result, the reliability of the PLSR model processed with the SG derivative method was the highest, and the root mean square average was the lowest among the three models. The predictability of the concentration of each element using the PLSR model pre-processed by the SG derivative was confirmed with the residual predictive deviation parameter. The predicted calorific value was estimated from the predicted concentrations of elements in coal using Dulong’s equation. The PLSR model pretreated by the SG derivative showed the lowest error compared to the calorific value of mixed coals obtained via the chemical analysis.