Fusion of Laser-Induced Breakdown and Near-Infrared Spectroscopies for Quantitative Detection of Industrial Indicators in Coal

To enhance the rapid detection capability of industrial coal indicators, this study proposes an analytical method based on the fusion of laser-induced breakdown spectroscopy (LIBS) and near-infrared (NIR) spectroscopy. Forty coal samples were collected, their LIBS and NIR spectra were preprocessed, and both single-source and spectral fusion models were constructed. Feature extraction was performed using Variable Importance in Projection (VIP) and the Least Absolute Shrinkage and Selection Operator (LASSO), followed by the establishment of quantitative prediction models via partial least squares regression (PLS) and support vector regression (SVR). The results show that feature-level fusion models outperform single-spectral source models. Specifically, the feature-level fusion model integrating LASSO feature extraction and PLS achieved the best comprehensive performance, with coefficients of determination for prediction (Rp2) of ash, volatile matter, and calorific value of 0.9622, 0.9687, and 0.9486, respectively, with corresponding root mean square errors of prediction of 2.47%, 1.77%, and 1.21 MJ/kg. These findings demonstrate that LIBS and NIR are complementary in providing information on elemental composition and molecular structure, and that spectral fusion effectively improves the prediction accuracy of coal quality indicators, offering a feasible approach for rapid coal quality analysis.