矿物化学成分的精确确定对矿产资源开发利用具有重要意义, 利用热红外光谱反演铁矿中SiO2含量弥补了传统方法耗时长等方面的不足。 然而铁矿的热红外光谱受表面粗糙度(roughness, Rq)等因素影响, 导致SiO2含量反演精度降低。 现有研究在没有考虑矿石表面粗糙度对成分反演影响的情况下, 利用热红外光谱数据对铁矿石中SiO2含量进行定量反演, 反演精度对精确圈定矿体范围及配矿难以提供有效帮助。 因此, 将粗糙度作为影响反演铁矿中SiO2含量的考虑因素, 研究对反演SiO2精度的影响。 以辽宁省的“鞍山式”铁矿为研究对象, 为满足热红外光谱观测要求, 将铁矿试样制备成直径6 cm、 厚度1 cm的圆柱形块体共14块, 按其SiO2含量多少形成序列。 每件试样正反两面制作成两个等级的粗糙度, 并利用Surtronic S128粗糙度仪观测表面粗糙度。 采用红外光谱辐射计Turbo FT观测试样热红外光谱发射率, 利用归一化指数(NDI)分析光谱指数与SiO2含量的相关性, 确定两个等级粗糙度试样SiO2含量的敏感波段分别位于8.12, 8.13和8.02, 8.03 μm处, 相关系数分别为0.947和0.972。 建立敏感波段与试样SiO2含量的定量反演模型, 分析粗糙度对反演SiO2含量的影响。 结果表明: (1)粗糙度Rq增加对RF(reststrahlen features)特征区域光谱发射率增加影响明显。 平均粗糙度Rq由1.05 μm增加到2.47 μm, 使得同一块试样粗糙面与光滑面发射率的最大差值为0.17(相对差42.9%)。 (2)相同等级粗糙度进行含量反演时, 反演误差小, 平均相对误差1.88%, 大部分试样的反演精度能够满足地质矿产行业标准的误差要求。 (3)实验结果较不考虑铁矿表面形态反演SiO2含量精度3.57%有较大提高, 相对提高精度为47.3%。 因此, 考虑粗糙度的影响对提高SiO2含量的反演精度, 实现铁矿的精确区划, 合理、 高效的开采铁矿资源具有重要意义。

The precise determination of mineral chemical composition is significance to the exploitation and utilization of mineral resources, and inversion of SiO2 content in iron ore by thermal infrared spectrum makes up for the shortcomings of the traditional methods in terms of time-consuming and so on. The thermal infrared spectrum of iron ore, however, is affected by surface roughness and other factors, which results in the decrease of the inversion accuracy of SiO2 content. The recent study doesn’t consider the influence of ore surface roughness on the inversion of ore composition and quantitatively inverted SiO2 content in iron ore by thermal infrared spectrum. The inversion result can’t provide any effective help for precise delineation of ore body range and ore blending. Therefore, this paper aims roughness on the factor to influence the inversion of SiO2 content in iron ore. Taking the “Anshan-type” iron ore in Liaoning Province as the research object, the samples are made into a total of 14 blocks of cylindrical blocks with a diameter of 6 cm and a thickness of 1 cm, which formed a sequence according to their SiO2 content. Two levels of roughnesses are made on both sides of each sample, and the surface roughness is observed by using Surtronic S128 roughness meter. The infrared spectroradiometer Turbo FT is used to observe the thermal infrared spectroscopy emissivities of samples. The correlation indexes between the spectral index and SiO2 content are analyzed by the normalized index (NDI) to determine the sensitive bands of SiO2 content of two grade roughness samples. Located at 8.12~8.13, 8.02~8.03 μm, the correlation coefficients are 0.947 and 0.972, respectively. A quantitative inversion model of the sensitive band and SiO2 content is established to analyze the effect of roughness on the inversion of SiO2 content. The results show that: (1) The increase of roughness Rq has a significant effect on the spectral emissivity of RF(Reststrahlen Features) characteristic regions. The average roughness Rq is increased from 1.05 to 2.47 μm, so that the maximum difference between the rough surface and the smooth surface emissivity of the same sample is 0.17 (relative difference 42.9%). (2) When the same grade roughnesses are used for content inversion, the inversion error is small, and the average absolute error is 1.88%. The inversion accuracy of most samples can meet the error requirements of the geological and mineral industry standards. (3) The experimental results of inversion SiO2 content accuracy are great higher than the inversion accuracy of 3.57% without considerating the iron ore surface morphology, and the relative improvement accuracy is 47.3%. Therefore, considering the influence of roughness is of great significance for improving the inversion accuracy of SiO2 content, then it is of great significance to realize the precise division of iron ore and mine iron ore resources reasonably and efficiently.