School of Chemical Engineering and Blasting, Anhui University of Science & Technology
Inner Mongolia Kangning Blasting Co. , Ltd.

为实现露天矿山台阶爆破效果预测,以爆破经验公式和现场爆破数据为基础,构建了爆破参数经验数据集,采用BP、FNN和RBF神经网络智能算法对爆破参数优化进行研究和分析。基于深度神经网络算法对爆破参数与岩石破碎的关系进行研究和分析,建立了爆破参数与大块率的预测模型,对爆破参数进行敏感性分析,并将预测结果与实例对比。研究结果表明:3种预测模型的训练过程的损失值均小于0.05,对敏感性分析发现,孔距和排距对模型预测结果具有最显著的影响;在数据集的训练和测试中,BP模型表现出优越的预测精度,FNN模型在各方面表现均衡,RBF模型表现出显著的稳定性;在应用实例中,3种模型的相对误差均不超过10%,在预测爆破结果参数方面具有较高准确度,此研究模型和结果可以作为爆破工程实践的参考。

To predict the blasting effect in open-pit mine bench blasting, an empirical data set of blasting parameters is constructed based on established blasting experience formulas and field blasting datas. Intelligent algorithms, including BP, FNN, and RBF neural networks, are used to research and analyze the optimization of blasting parameters. Based on a deep neural network algorithm, the relationship between blasting parameters and rock fragmentation is examined, leading to the establishment of a predictive model for blasting parameters and large block rate. Additionally, a sensitivity analysis of the blasting parameters is conducted, and the prediction results are compared with practical examples. The research results indicate that the loss values of the three models during training process are all below 0. 05. The sensitivity analysis reveals that hole spacing and row spacing exert the most significant influence on the model prediction results. In the training and testing phases of the data set, BP model demonstrates superior prediction accuracy, while the FNN model exhibits balanced performance in all aspects. Additionally, RBF model displays notable stability. In practical applications, the relative errors Coal Mine Blasting Vol. 42 No. 4 Dec. 2024 of the three models do not exceed 10%, confirming their high accuracy in predicting blasting parameters. The models and results presented in this article can provide a reference for blasting engineering practice.

安徽高校自然科学研究项目(2022yjrc17);安徽理工大学高层次引进人才科研启动基金(2022AH050838);安徽理工大学大学生创新创业训练计划项目(202310361042)

戴增杰,梁昊,王贵,等.基于3种神经网络算法的露天矿山台阶爆破块度预测[J].煤矿爆破,2024,42(4):1-6.

DAI Zengjie,LIANG Hao,WANG Gui,et al. Block size prediction for bench blasting in open-pit mine based on three neural network algorithms[J]. Coal Mine Blasting,2024,42(4):1-6.