摘要
由于矿石矿物的分布范围广,矿石数量巨大,使矿石矿物难以被识别。矿石矿物的形成类型受外力影响,导致矿体中不同部位的构成元素不同,而形成的矿石矿物种类不同。某些矿石中适用于工业生产的元素较少,在开采的过程中会导致回收率低,不能为工业所利用。这种矿石矿物的开采浪费人力物力,使用智能矿石矿物智能识别算法识别出适合开采的矿石矿物将有助于提高矿石矿物开采利润。使用基于CART和PU学习算法的矿石矿物的智能识别,研究适合开采矿石矿物的识别问题,首先从获取的数据中进行样本制作,然后使用PU学习算法针对数据中没有负向样本标注的问题进行负向样本标注,得到完整的样本数据。最后使用样本数据对CART算法进行训练,得出CART算法分类器。通过实验得到基于CART算法和PU学习算法模型的准确率为89.45%,对比ID3算法和C4.5算法得到较为准确的识别结果。
Due to the wide distribution of ore minerals,the amount of ore is huge,the structure is complex due to the influence of external forces,and the elements in different parts of the ore body are different,different mineral ore types are formed.Some ore have fewer elements suitable for industrial production,which will lead to a low recovery rate during the mining process.This kind of can not be used by industry,wasting labor and material resources.Using intelligent identification of ore minerals suitable for mining intelligently will help to increase the profit of ore mineral mining.This article uses the intelligent identification of ore minerals based on CART and PU learning algorithms to study the identification of mining ore minerals.First,samples are extracted from the acquired data for sample production,and then the PU learning algorithm is used for the data without negative sample labeling in the data.Then,PU learning algorithm is used to conduct negative sample labeling for the problem that there is no negative sample labeling in the data to obtain positive and negative samples.Finally,the CART algorithm is trained using positive and negative samples to get the CART algorithm classifier.The accuracy rate of the model based on the CART algorithm and the PU learning algorithm was 89.45%through experiments,compared with the ID3 algorithm and the C4.5 algorithm,it obtained more accurate recognition results.
作者
赵永翼
申莹
王菲
ZHAO Yongyi;SHEN Ying;WANG Fei(Software College,Shenyang Normal University,Shenyang 110034,China)
出处
《沈阳师范大学学报（自然科学版）》
CAS
2020年第2期176-182,共7页
Journal of Shenyang Normal University:Natural Science Edition
基金
辽宁省教育厅高等学校基本科研项目(LFW201702)。
关键词
CART算法
矿物自动识别
智能地质学
数据挖掘
Cart algorithm
mineral automatic recognition
intelligent geology
data mining
