摘要
为了更好地预测岩矿石爆破块度,将世界多个矿山的爆破块度统计数据依据弹性模量分成两组,运用随机森林和GA-BP神经网络分别建立爆破数据随机森林分组模型和爆破块度预测模型。以孔间间距与排间间距比、炮孔孔深与排间间距比、排间间距与炮孔直径、炮孔堵塞长度与排间间距比、炸药单耗、原位岩石块度和岩石弹性模量为输入参数,对爆破块度进行预测。结果表明预测结果的相关性系数(R2)、均方根误差(RMSE)和平均相对误差(MRE)均优于多元回归预测模型和BP神经网络预测模型,且优于未分组情况下建立的GA-BP神经网络预测模型,更适合于爆破工程的实际应用,并为多因素影响下的爆破块度预测提供了一种新思路。
In order to better predict the blasting fragmentation,the blasting fragmentation statistics data of multiple mines in the world were divided into two groups.Random forest grouping model for blasting data and prediction model of blasting fragmentation were established by random forest and GA-BP neural network.The ratio of hole spacing to row interval,ratio of bench height to row interval,ratio of row interval to hole diameter,ratio of stemming to row interval,powder factor,in situ block size and elasticity modulus were taken as input parameters,and the blasting fragmentation was predicted.The results show that the correlation coefficient(R2),root mean square error(RMSE),and mean relative error(MRE)of the prediction results are better than those of the multiple regression prediction model and BP neural network prediction model,and better than those of the GA-BP neural network prediction model established without grouping,which are more suitable for the practical application of blasting operation.The study provides a new idea for the prediction of blasting fragmentation under the influence of multiple factors.
作者
刘阳
谭凯旋
郭钦鹏
王鹏
何成垚
LIU Yang;TAN Kaixuan;GUO Qinpeng;WANG Peng;HE Chengyao(School of Resourse Environment and Safety Engineerings University of South China,Hengyang,Hunan,421001,China;School of Mathematics and Physics,University of South China,Hengyang,Hunan 421001,China;Hunan Key Laboratory of Rare Metal Minerals Exploitation and Geological Disposal of Waste,Hengyang,Hunan 421001,China)
出处
《矿业研究与开发》
CAS
北大核心
2021年第1期135-140,共6页
Mining Research and Development
基金
国家自然科学基金项目(U1703123)。
