叶海旺,胡俊杰,雷涛,李宁,王其洲,DAHé MONYEGNI GHISLAIN

• 1:武汉理工大学资源与环境工程学院
• 2:矿物资源加工与环境湖北省重点实验室
• 3:武汉理工大学安全科学与应急管理学院

摘要(Abstract)：

针对小样本条件下使用机器学习方法预测岩石爆破块度存在的数据利用率不足，预测精度存在浮动较大的问题。将留一法(LOO)与极限梯度提升(XGboost)算法结合，利用Python 3.7构建LOO-XGboost岩石爆破块度预测模型，选取31组爆破数据进行LOO-XGboost模型训练与预测，通过调用不同参数，得到模型的最优内置参数如下：求解方式为树模型，学习率为0.30,决策树数量为50,决策树最大迭代深度为3,叶子节点最小样本数为3,随机采样比例为0.8。将预测结果与同条件下的支持向量机回归(SVR)、BP神经网络(BPNN)、随机森林(RF)模型以及10折交叉验证下的XGboost模型进行对比，对比结果为LOO-XGboost模型预测精度明显高于其他4种模型，其相关系数、均方根误差、平均绝对误差分别为0.9128、0.0587、0.0342。结果表明：LOO-XGboost模型既保证了小样本情况下的数据利用率，又提高了预测精度，适合应用于岩石爆破块度预测。

关键词(KeyWords)： 岩石块度;LOO-XGboost模型;小样本预测;模型对比;Python 3.7

Abstract:

Keywords:

基金项目(Foundation): 国家重点研发计划(2020YFC1909602,2021YFC2902901);; 湖北省重点研发计划(2021BCA152)

作者(Author): 叶海旺,胡俊杰,雷涛,李宁,王其洲,DAHé MONYEGNI GHISLAIN

参考文献(References)：

• [1] GHEIBIE S,AGHABABAEI H,HOSEINIE S H,et al.Modified Kuz-Ram fragmentation model and its use at the Sungun Copper Mine[J].Int J Rock Mech Min Sci,2009,46(6):967-973.
• [2] 聂军，史秀志，陈新，等.基于GEP的露天矿台阶爆破块度预测模型[J].爆破，2015,32(2):82-88.[2] NIE Jun,SHI Xiu-zhi,CHEN Xin,et al.Prediction model of bench blasting fragmentation based on GEP in open-pit mine[J].Blasting,2015,32(2):82-88.(in Chinese)
• [3] 史秀志，王洋，黄丹，等.基于LS-SVR岩石爆破块度预测[J].爆破，2016,33(3):36- 40.[3] SHI Xiu-zhi,WANG Yang,HUANG Dan,et al.Based on LS-SVR rock blasting fragmentation prediction[J].Blasting,2016,33(3):36- 40.(in Chinese)
• [4] 史秀志，周健，吴帮标，等.台阶爆破岩石破碎平均粒径预测的支持向量机方法(英文)[J].Transactions of Nonferrous Metals Society of China,2012,22(2):432- 441.[4] SHI Xiu-zhi,ZHOU Jian,WU Bang-biao,et al.Support vector machine method for predicting average particle size of rock broken by bench blasting(English)[J].Transactions of Nonferrous Metals Society of China,2012,22(2):432- 441.(in Chinese)
• [5] 刘阳，谭凯旋，郭钦鹏，等.运用随机森林和GA-BP神经网络预测岩石爆破块度[J].矿业研究与开发，2021,41(1):135-140.[5] LIU Yang,TAN Kai-xuan,GUO Qin-peng,et al.Using random forest and GA-BP neural network to predict rock blasting fragmentation[J].Mining Research and Development,2021,41(1):135-140.(in Chinese)
• [6] 唐跃，徐曲，柯波，等.基于交叉验证的矿岩爆破块度SVM模型优选研究[J].爆破，2018,35(3):74-79.[6] TANG Yue,XU Qu,KE Bo,et al.Optimization of SVM model of rock fragmentation based on cross-validation[J].Blasting,2018,35(3):74-79.(in Chinese)
• [7] 柳小波，袁鹏喆，张兴帆.基于RBF神经网络的露天矿爆破效果预测研究[J].中国矿业，2020,29(1):81-84.[7] LIU Xiao-bo,YUAN Peng-zhe,ZHANG Xing-fan.Research on the prediction of blasting effect of open-pit mine based on RBF neural network[J].China Mining,2020,29(1):81-84.(in Chinese)
• [8] 潘玉忠，张义平，王强，等.台阶爆破块度的SVM预测模型研究[J].矿业研究与开发，2010,30(5):97-99.[8] PAN Yu-zhong,ZHANG Yi-ping,WANG Qiang,et al.SVM prediction model of bench blasting fragmentation[J].Mining Research and Development,2010,30(5):97-99.(in Chinese)
• [9] 王仁超，朱品光.基于随机森林回归方法的爆破块度预测模型研究[J].水力发电学报，2020,39(1):89-101.[9] WANG Ren-chao,ZHU Pin-guang.Study on prediction model of blasting fragmentation based on random forest regression method[J]Journal of Hydroelectric Engineering,2020,39(1):89-101.(in Chinese)
• [10] CHEN Tian-qi,CARLOS Guestrin.XGBoost:a scalable tree boosting system.[J].IEICE Transactions on Fundamentals of Electronics,Communications and Computer Sciences,2016,abs/1603.02754.
• [11] 李晨阳，陈雄飞，张勇，等.基于XGBoost的铝合金LIBS光谱分类识别方法[J].光谱学与光谱分析，2021,41(2):624- 628.[11] LI Chen-yang,CHEN Xiong-fei,ZHANG Yong,et al.Classification and identification method of aluminum alloy LIBS spectrum based on XGBoost[J].Spectroscopy and Spectral Analysis,2021,41(2):624- 628.(in Chinese)
• [12] 闫星宇，顾汉明，肖逸飞，等.XGBoost 算法在致密砂岩气储层测井解释中的应用[J].石油地球物理勘探，2019,54(2):447- 455.[12] YAN Xing-yu,GU Han-ming,XIAO Yi-fei,et al.Application of XGBoost algorithm in logging interpretation of tight sandstone gas reservoirs[J].Petroleum Geophysical Prospecting,2019,54(2):447- 455.(in Chinese)
• [13] 谢学斌，李德玄，孔令燕，等.基于CRITIC-XGB算法的岩爆倾向等级预测模型[J].岩石力学与工程学报，2020,39(10):1975-1982.[13] XIE Xue-bin,LI De-xuan,KONG Ling-yan,et al.Rockburst tendency grade prediction model based on CRITIC-XGB algorithm[J].Chinese Journal of Rock Mechanics and Engineering,2020,39(10):1975-1982.(in Chinese)
• [14] 张福浩，朱月月，赵习枝，等.地理因子支持下的滑坡隐患点空间分布特征及识别研究[J].武汉大学学报(信息科学版),2020,45(8):1233-1244.[14] ZHANG Fu-hao,ZHU Yue-yue,ZHAO Xi-zhi,et al.Research on spatial distribution characteristics and identification of landslide hidden danger points supported by geographical factors[J].Journal of Wuhan University(Information Science Edition),2020,45(8):1233-1244.(in Chinese)
• [15] 闫星宇，顾汉明，肖逸飞，等.XGBoost算法在致密砂岩气储层测井解释中的应用[J].石油地球物理勘探，2019(2):447- 455.[15] YAN Xing-yu,GU Han-ming,XIAO Yi-fei,et al.Application of XGBoost algorithm in logging interpretation of tight sandstone gas reservoirs[J].Petroleum Geophysical Prospecting,2019(2):447- 455.(in Chinese)
• [16] HUDAVERDI T,KULATILAKE P H S W,KUZU C.Prediction of blast fragmentation using multivariate analysis procedures[J].Int J Numer Anal Methods Geomech,2011,35(12):1318-1333.
• [17] 王晓晖，张亮，李俊清，等.基于遗传算法与随机森林的XGBoost改进方法研究[J].计算机科学，2020,47(S2):454- 458,463.[17] WANG Xiao-hui,ZHANG Liang,LI Jun-qing,et al.Research on improved method of XGBoost based on genetic algorithm and random forest[J].Computer Science,2020,47(S2):454- 458,463.(in Chinese)