摘要
为了准确快速地确定充填体强度,依据试验数据,以料浆体积分数、水泥掺量、人工砂尾砂比以及养护时间作为输入因子,以充填体的单轴抗压强度作为输出因子,建立了一种充填体强度ANN-PSO预测模型。研究结果表明,该模型的预测性能较好,在预测充填体强度时其平均相对误差率MAP为2.41%,可决系数R^2为0.983。通过对比136组充填配合比充填体的室内试验强度值和实际生产测定值,获得了两者之间的强度折减系数。利用预测模型并联合强度折减系数,预测得到了矿山运行期间160多条进路的充填体强度值。该模型可大幅减小物理试验量,为类似的充填矿山提供了良好的借鉴作用。
In order to accurately and quickly determine the strength of filling body,a strength prediction model of filling body based on ANN-PSO was established according to experimental data.The slurry density,cement dosage,ratio of artificial aggregate and tailings,as well as curing time were taken as input factors,and uniaxial compressive strength of filling body was taken as output factor.The results showed that the model had a good prediction performance for the strength of filling body.When predicting the filling strength,the average relative error rate MAP was 2.41% and the determination coefficient R^2 was 0.983.By comparing the experimental strength values of 136 groups of filling bodies and the measured values of actual production,the strength reduction coefficient between them was obtained.By using this model and combining the strength reduction coefficient,the strength values of filling bodies in more than 160 routes during the running period were predicted.It can greatly reduce the amount of physical experiments,which provided a good reference for similar backfill mines.
作者
吴炜
吉坤
张朋
邱剑辉
骆禧光
韩斌
WU Wei;JI Kun;ZHANG Peng;QIU Jianhui;LUO Xiguang;HAN Bin(Guizhou Jinfeng Mining Co.,Ltd,Guiyang,Guizhou 530002,China;School of Civil and Resource Engineering,University of Science and Technology Beijing,Beijing 100083,China)
出处
《矿业研究与开发》
CAS
北大核心
2020年第2期53-57,共5页
Mining Research and Development
基金
国家自然科学基金资助项目(51304011)
关键词
充填体强度
预测
人工神经网络
粒子群算法
强度折减
Strength of filling body
Prediction
Artificial neural network
Particle swarm optimization
Strength reduction
