基于改进随机森林算法的岩石爆破块度预测  

Prediction for Blasting Fragmentation of Rocks Based on Improved Random Forest Regression Method

作　　者：刘翔 谢涛 王训洪 易泽邦 何秋芝 LIU Xiang;XIE Tao;WANG Xunhong;YI Zebang;HE Qiuzhi(Guangxi University of Science and Technology,Liuzhou,Guangxi 545006,China;Liuzhou Weiyu Blasting Engineering Co.,Ltd,,Liuzhou,Guangxi 545002,China;The Eleventh Metallurgical Construction Group Co.,Ltd.,Liuzhou,Guangxi 545027,China;Guilin University of Technology,Guilin,Guangxi 541004,China)

机构地区：[1]广西科技大学,广西柳州市545006 [2]柳州威宇爆破工程有限责任公司,广西柳州市545002 [3]十一冶建设集团有限责任公司,广西柳州市545027 [4]桂林理工大学,广西桂林市541004

出　　处：《矿业研究与开发》2022年第7期25-29,共5页Mining Research and Development

基　　金：国家自然科学基金青年基金项目(42003066);广西科技基地与人才专项项目(2021AC19198,2021AC19200);广西科技大学博士基金项目(校科博20S10,21Z29)。

摘　　要：为提高岩石爆破块度预测效果,利用多个矿山的岩石爆破统计数据,通过影响爆破岩石块度因素的重要度计算和皮尔逊相关系数判定筛选出炸药单耗、岩石块度尺寸、岩石弹性模量以及炮孔堵塞长度与炮孔排距比(T/B)等6个特征变量作为输入参数,建立一种基于改进随机森林回归算法的爆破块度预测模型。该模型预测的爆破块度逼近真实值,预测结果的可决系数(R)、均方根误差(RMSE)和平均相对误差(MRE)分别为0.9881,0.0430和0.1445,相较于线性回归预测模型和BP神经网络预测模型而言,其预测效果更优,因此该模型在实际应用中更具适用性,能够为爆破参数设计和优化提供参考。In order to improve the prediction effect of blasting fragmentation of rocks,the rock blasting data from several mines was used.By calculating the importance degree of the factors affecting the blasting rock fragmentation and judging the Pearson correlation coefficient,six characteristic variables such as explosive unit consumption,rock fragmentation size,rock elastic modulus,blast hole plugging length and blast hole row spacing ratio were selected as input parameters,and a blasting fragmentation prediction model based on improved random forest regression algorithm was established.The blasting fragmentation predicted by the model is close to the real value,and the determinable coefficient(R~2),root mean square error(RMSE)and average relative error(MRE)of the predicted results are 0.9881,0.0430 and 0.1445 respectively.Compared with linear regression prediction model and BP neural network prediction model,the prediction effect of the model is better.Therefore,the model is more available in practical application and can provide a reference for the design and optimization of blasting parameters.

关 键 词：爆破块度 随机森林回归模型 块度预测 预测精度 

分 类 号：TD235[矿业工程—矿井建设] TD804