X射线荧光光谱结合判别分析识别进口铁矿石产地及品牌

X-Ray Fluorescence Spectroscopy Combined With Discriminant Analysis to Identify Imported Iron Ore Origin and Brand

铁矿石是钢铁工业的重要原材料,不同产地、品牌的进口铁矿石在元素组成、含量上存在差异,进口铁矿石掺杂、掺假、以次充好等现象虽集中于个案,却危害经济安全。故建立主要进口国铁矿石产地与品牌的快速识别模型,对支撑进口铁矿石的风险监管,保障贸易便利化。该研究对象为澳大利亚、南非、巴西3个国家共14个品牌的236份进口铁矿石样品,包括皮尔巴拉混合粉(块)、杨迪粉铁矿,纽曼混合粉(块)铁矿、津布巴混合粉铁矿、国王粉、弗特斯克混合粉、昆巴标准粉(块)、卡拉加斯铁矿石等。应用波长色散-X射线荧光光谱无标样分析法测定所有研究样品的元素组成及含量,检出元素包括Fe, O, Si, Ca, Al, Mn, Tb, Ti, Mg, P, K, S, Cr, Na, Sr, Zr, Zn, V, Cu, Gd, Ba, Cl, Ni和Co,共计24种,选择其中Fe, O, Si, Ca, Al, Mn, Tb, Ti, Mg, P, Cr和S共12种所有样品全部检出的元素进行判别分析。采用逐步判别法筛选出Fe, O, Si, Ca, Al, Mn, Ti, Mg, P和S共10个元素含量作为有效变量,建立二维Fisher判别模型,实现对澳大利亚、南非、巴西进口铁矿石的识别,模型对建模样品识别正确率为97.40%,交叉验证正确率为95.30%,对测试样品的识别正确率达到95.50%。针对14种品牌铁矿石,使用Fe, O, Si, Ca, Al, Mn, Ti, Mg, P和S共10种元素含量,建立十维Fisher判别模型,模型对建模样品识别正确率为100%,交叉验证正确率为97.90%,对测试样品的识别正确率达到100%。波长色散-X射线荧光光谱无标样分析虽然是一种半定量分析方法,但分析快速,稳定性好,该方法结合逐步判别-Fisher判别分析,能实现对铁矿石产地与品牌的识别。

Iron ore is an important raw material for the iron and steel industry. Imported iron ore with different origins and brands varies in elemental composition and content. Phenomena such as doping, adulteration and shoddy of imported iron ore are endangering the national security and economy safety, so it is necessary to establish a rapid identification model of the origin and brand of imported iron ore in major importing countries, can support the risk supervision of imported iron ore, and ensure trade facilitation. The research objects of this paper are 236 imported iron ore samples from 14 brands in Australia, South Africa and Brazil, including Pilbara Blend Fines(Lumps), Yandi Fines, Newman Blend Fines(Lumps), Jimblebar Blend Fins, Kings Fines, Fortescue Blend Fines, Kumba Standard Fines(Lumps), and Carajas Iron Ore, etc. The elemental composition and content of all research samples were determined by wavelength dispersive X-ray fluorescence spectrum standard-less analysis method, and it turned out that elements detected from iron ore samples are 24 in total, including Fe, O, Si, Ca, Al, Mn, Tb, Ti, Mg, P, K, S, Cr, Na, Sr, Zr, Zn, V, Cu, Gd, Ba, Cl, Ni, and Co. Among them, we chose 12 elements and conducted a stepwise discriminant-Fisher discriminant analysis modeling, including Fe, O, Si, Ca, Al, Mn, Tb, Ti, Mg, P, Cr, and S. Moreover, 10 elements including Fe, O, Si, Ca, Al, Mn, Ti, Mg, P, S were screened out as valid variables by the stepwise discriminant method. A two-dimensional Fisher discriminant model was thus established to realize the identification of imported iron ore from Australia, South Africa and Brazil. The recognition accuracy of the model for the modeled sample was 97.40%, the one of cross-validation was 95.30%, and that of the test sample reached 95.50%. For the 14 brands of iron ore, 10 elements including Fe, O, Si, Ca, Al, Mn, Ti, Mg, P, and S were used to establish a ten-dimensional Fisher discriminant model, and its recognition accuracy for the modeled sample was 100%. The accuracy of cross-validation was 97.90%, while one of the test samples reached 100%. Although wavelength dispersion X-ray fluorescence spectrum standard-less analysis method is a semi-quantitative analysis method, the analysis is fast and stable, wavelength dispersive X-ray fluorescence spectrum standard-less analysis method together with the stepwise discriminant-Fisher discriminant analysis can realize the identification of importing countries and brands of iron ore.