基于XGBoost的矿石加工合格率预测研究  

Research on prediction model of ore processing qualification rate based on XGBoost

作　　者：李凝思 王娇娅 谷佳妍 赵嘉豪 李东[4] LI Ningsi;WANG Jiaoya;GU Jiayan;ZHAO Jiahao;LI Dong(Department of Electronic Engineering,Taiyuan Institute of Technology,Taiyuan 030008,China;Department of Computer Engineering,Taiyuan Institute of Technology,Taiyuan 030008,China;Department of automation,Taiyuan Institute of Technology,Taiyuan 030008,China;School of Software Engineering,South China University of Technoly,Guangzhou 510006,China)

机构地区：[1]太原工业学院电子工程系,太原030008 [2]太原工业学院计算机工程系,太原030008 [3]太原工业学院自动化系,太原030008 [4]华南理工大学软件学院,广州510006

出　　处：《计算机应用文摘》2022年第21期92-94,共3页Chinese Journal of Computer Application

摘　　要：矿石加工的过程十分复杂,多种因素(特别是温度)会对矿石加工的最终质量产生重要影响。文章利用2022年“五一杯数模竞赛”给出的原始数据集和矿石加工过程进行系统建模,并通过指定系统Ⅰ和Ⅱ的设定温度来对矿石加工的产品合格率进行预测。为了建立合适的模型,文章选择了XGBoost回归、随机森林回归、SVM回归、EP神经网络回归等四种机器学习方法建立预测模型,通过对比回归分析的拟合程度R2,最终选择XGBoost对矿石加工过程的数据进行建模,并基于该模型进行产品合格率预测。The process of ore processing is very complex,and many factors(especially temperature)will have an important impact on the final quality of ore processing.In this paper,the original data set and ore processing process given by the 2022"May 1st Cup Mathematical Contest"are used for system modeling,and the product qualification rate of ore processing is predicted by specifying the set temperature of system I and II.In order to establish a suitable model,this paper selects four machine learning methods,namely XGBoost regression,random forest regression,SVM regression and BP neural network regression,to establish a prediction model.By comparing the fitting degree R2 of regression analysis,XGBoost is finally selected to model the data of ore processing process and predict the product qualification rate based on this model.

关 键 词：矿石加工 产品合格率 预测 XGBoost 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]