Fast measurement of coking properties of coal using laser induced breakdown spectroscopy

Abstract

Fast measurement of caking index (G) and maximum thickness of plastic layer (Y) of coal is urgently needed for process optimization in coking industry to improve coke quality and reduce CO2 and pollution emission. Till now, there has not been any report on fast determination of these two coal property parameters yet. In the present work, laser-induced breakdown spectroscopy (LIBS) was utilized for the fast measurement of G and Y value as the first time. A dominant factor based partial least square (PLS) model combined with variable selection was applied to improve measurement performance using 189 coal samples. Thirteen important lines which physically correlate with G and Y were chosen in establishing the dominant factor model, and the residual errors were further corrected by PLS together with variable selection to avoid noise overfitting. The proposed model showed satisfactory quantitative measurement of G and Y, which were very close to the Chinese national standard based on traditional laboratory method. Compared with the traditional PLS model, the root mean square error of prediction (RMSEP) for G value was reduced from 5.117 to 4.197, with an improvement of 17.9% relatively. The RMSEP for Y value was reduced from 1.588 to 1.036, with an improvement of 34.7% relatively. The improvement brought by the dominant factor model and the variable selection method was discussed in detail. Results showed that LIBS is capable of fast measurement of the values of G and Y of coal for coke industry.