Bai Yang, Xie Xiaoxiao, Jia Yuna, Zhang Yunpeng
College of Mining Engineering, North China University of Science and Technology, Tangshan, 063210, China.
School of Emergency Management and Safety Engineering, North China University of Science and Technology, Tangshan, 063210, China.
Sci Rep. 2025 Aug 13;15(1):29699. doi: 10.1038/s41598-025-15429-3.
In order to clarify the relationship between magnetite grade and reflectance spectra, spectral testing of different grades of magnetite was carried out to analyze the effect of iron grade on the reflectance spectra of specimens. The results show that the fluctuation trends of the spectral curves of different grades of magnetite are slightly different. By screening the characteristic bands of magnetite grade, it was found that the characteristic bands of different grades of magnetite are mainly distributed in the range of 350-1000 nm and 1800-2500 nm. Then based on the relationship between magnetite grade and reflectance spectra, two inversion models of magnetite grade, PLSR(Partial Least Squares Regression) and MLSR(Multiple Linear Stepwise Regression), were established. By comparing and analyzing the inversion models of magnetite grade established by PLSR and MLSR, it was found that the inversion accuracy based on the PLSR model was better than that of the MLSR model, in which the [Formula: see text] and the RMSE of validation set of the best model was respectively 0.9982 and 0.95%.
为了阐明磁铁矿品位与反射光谱之间的关系，对不同品位的磁铁矿进行了光谱测试，以分析铁品位对样品反射光谱的影响。结果表明，不同品位磁铁矿的光谱曲线波动趋势略有不同。通过筛选磁铁矿品位的特征波段，发现不同品位磁铁矿的特征波段主要分布在350 - 1000nm和1800 - 2500nm范围内。然后基于磁铁矿品位与反射光谱的关系，建立了两种磁铁矿品位反演模型，即偏最小二乘回归（PLSR）模型和多元线性逐步回归（MLSR）模型。通过对PLSR和MLSR建立的磁铁矿品位反演模型进行比较分析，发现基于PLSR模型的反演精度优于MLSR模型，其中最佳模型的决定系数[公式：见原文]和验证集均方根误差分别为0.9982和0.95%。
