Hyperspectral analysis and inversion model study of water content in magnetite

Excessive moisture in iron ore hinders pulverization, beneficiation, and smelting. Traditional detection methods struggle with complex ore compositions. This study developed a hyperspectral model to estimate moisture in Hebei magnetite. Samples were preprocessed using S-G smoothing, MSC, SNV, derivatives, and continuum removal. CARS selected optimal bands, while PSO-LSSVR improved modeling. Reflectance showed a negative correlation with moisture, with Fe³⁺ (990 nm) and -OH (1440/1920 nm) absorption features. The models achieved R² values of 0.798 and 0.648. The integrated approach outperformed conventional methods, enabling accurate moisture detection for industrial use.