Determination of coal quality plays a major role in coal-fired power plants and coal producers for optimizing the utilization efficiency and controlling the quality. In this work, a rapid coal analyzer based on laser-induced breakdown spectroscopy (LIBS) was developed for rapid quality analysis of pulverized coal. The structure of the LIBS apparatus was introduced in detail. To avoid time-consuming and complicated sample preparation, a pulverized feeding machine was designed to form a continuously stable coal particle flow. The standard deviation (SD) of characteristic peaks was used to estimate the spectral valid data in this experiment. Coupled with cluster analysis, artificial neural networks and genetic algorithm are employed as a nonlinear regression method in order to indicate the relationship between coal quality and the corresponding plasma spectra. It is shown that the average absolute error of ash, volatile matter, fixed carbon, and gross calorific value for the validation set is 0.82%, 0.85%, 0.96%, and 0.48 MJ/kg. The average standard deviation of repeated samples is 1.64%, 0.92%, 1.08%, and 0.86 MJ/kg, showing a high sample-to-sample repeatability. This rapid coal analyzer is capable of performing reliable and accurate analysis of coal quality.
Keywords: ANN; LIBS; Laser-induced breakdown spectroscopy; artificial neural networks; calorific value; coal proximate analysis; particle flow.