摘要
为提高露天矿抛掷爆破效果预测精度,进而反馈优化爆破参数设计。建立HHO-LSSVM(哈里斯鹰算法优化最小二乘支持向量机)模型,预测抛掷爆破效果;将该模型所得预测精度及效率与未经优化的LSSVM(最小二乘支持向量机)、ELM(极限学习机)、GA-BP(遗传算法优化BP神经网络)、PSO-LSSVM(粒子群算法优化最小二乘支持向量机)模型进行对比。研究结果表明:采用HHO-LSSVM模型相较于未经优化的LSSVM和ELM模型所得到的有效抛掷率、松散系数、最远抛掷距离的预测精度均具有更高的决定系数值,更小的均方根误差值;HHO-LSSVM模型预测的有效抛掷率、松散系数、最远抛掷距离与实测数据之间的平均误差分别为2.7015%,2.9834%,2.8345%,均在5%以内,说明HHO-LSSVM模型对抛掷爆破效果具有较好的预测精度。研究结果可为通过准确预测爆破效果进而反馈抛掷爆破的优化设计提供一定参考。
To improve the prediction accuracy of the blast casting effect in the open-pit mines,and then feedback and optimize the design of blasting parameters,a model for the prediction of blast casting effect by Harris Eagle algorithm(HHO)optimization of least squares support vector machine(LSSVM)was established,and the prediction accuracy and efficiency of this model were compared with those of unoptimized LSSVM and extreme learning machine(ELM),genetic algorithm optimization BP neural network(GA-BP)and particle swarm optimization LSSVM(PSO-LSSVM)models.The results show that compared with the unoptimized LSSVM and ELM models,all the prediction accuracies of the effective casting rate,loose coefficient and the farthest casting distance obtained by the HHO-LSSVM model have higher determination coefficient values and smaller root mean square error values.The average errors of the effective casting rate,loose coefficient and the farthest casting distance predicted by the HHO-LSSVM model and the measured data are 2.7015%,2.9834%and 2.8345%,respectively,all within 5%,indicating that the HHO-LSSVM model has good prediction accuracy for the blast casting effect.The research results can provide certain reference method for the optimal design of blast casting by accurately predicting the blasting effect and then feedback.
作者
李天翔
李江
卢亚峰
LI Tianxiang;LI Jiang;LU Yafeng(Cathay Safety Technology Co.,Ltd.,Beijing 100012,China;Wuhai Energy Bureau,Wuhai Inner Mongolia 016000,China)
出处
《中国安全生产科学技术》
CAS
CSCD
北大核心
2023年第S01期110-116,共7页
Journal of Safety Science and Technology
基金
国家重点研发计划项目(2021YFC3001905)
中国安全生产科学研究院基本科研业务费专项资金项目(2023JBKY04)
