Quantitative analysis of coal quality by mutual information-particle swarm optimization (MI-PSO) hybrid variable selection method coupled with spectral fusion strategy of laser-induced breakdown spectroscopy (LIBS) and fourier transform infrared spectroscopy (FTIR)

Abstract

Coal quality analysis can provide basic work for the design and operation of related equipment and technological processes, which mainly including ash content and volatile matter. Variable selection plays an important role in the model and properly variable selection methods which can promote the predictive performance of the model and reduce the modeling time. In this article, a hybrid variable selection method mutual information-particle swarm optimization (MI-PSO) was used to realize precise screening of laser-induced breakdown spectroscopy (LIBS) and fourier transform infrared spectrometer (FTIR) spectral characteristic variables. Firstly, to obtain the fusion spectrum matrix, the pre-processed LIBS and FTIR spectrum matrix were directly connected end to end. Then the MI was used to eliminate redundant variables in the spectral data, and the PSO was used to further filter the retained variables to find a set of variables with higher prediction accuracy. Finally, a mutual information-particle swarm optimization-kernel extreme learning machine (MI-PSO-KELM) model is constructed based on the selected variables that the ash content and the volatile matter variables are 6499, 5093 respectively. The performance of MI-PSO-KELM is better than the KELM model based on LIBS and FTIR and the KELM model based on LIBS and FTIR primary spectral fusion, and can more accurately predict the ash content and volatile matter of coal quality analysis. The results for the determination of ash content, the optimal cross-validation (root mean square error of cross validation (RMSECV) =1.6170, determination coefficient of cross validation (RCV2) =0.9684) and prediction set the root mean square error of prediction (RMSEP) =0.9687, determination coefficient of prediction (RP2 = 0.9821) results are obtained by the MI-PSO based on primary spectral fusion model. For the determination of the volatile matter, the best results of cross-validation (RMSECV = 1.2886, RCV2=0.9868) and prediction set (RMSEP = 1.3218, RP2=0.9789) are obtained by the MI-PSO based on primary spectral fusion model. It shows that the spectral fusion technology can help improve the performance of LIBS spectral quantitative analysis, and a spectrum fusion method based on LIBS and FTIR is established to realize the rapid determination of ash content and volatile matter of coal.