模式识别在采矿中的应用

Application of pattern recognition for the mining

1 非线性映照法识别原理1.1特征向量选取特征向量的选取是模式识别工作的关键,首先,要求选取对分类贡献大而较少的(不破坏分类情况的最少特征量数)特征量数目,再者要求样本数和空间维数的比值要大于3.为此我们从表1的七个特征量中根据加权因子的大小作了初选。再据此通过各种组合逐步降维进行筛选.由七个特征量中筛选出分类效果最好的4个固有特征向量,即铅、锌、氧化钙+氧化镁和二氧化硅的分析数据.

The method of pattern recognition applied to mining is recomendcd.Discriminating ore From the rock is carried out with Nonlinear mapping method of pattern recognition. Some chemical analytical data,such as Lead(Pb),Zinc(Zn),Molybdenum(Mo),Cop- per(Cu)and Silica(SiO_2)et al.,are chosen as characteristic vectors for classification.The results show that the discrimination between ore and rock is quite satisfactory.