基于BP神经网络的砂泥岩互层地质爆破参数优化研究

砂岩、泥岩互层地质在黄土地区路基工程中比较常见,其软硬相间的特性使得爆破参数设计困难,工程实践中难以达到理想的爆破效果。本文提出一种基于BP神经网络的爆破参数优化方法,该方法以影响爆破效果的主要因素作为网络输入参数、以爆破效果作为输出参数、采用搜集的数据样本作为训练和检测样本建立BP神经网络。以通过理论计算得出的爆破参数为基础,利用正交试验法筛选出16种试验方案。基于训练好的神经网络预测爆破效果,从中选择爆破效果较好的5组方案进行试爆,并根据试爆效果确定最佳爆破方案。该方法应用于横山车站路堑工程施工,效果良好。

Study on Blasting Parameters Optimization of Sandstone-mudstone Interbedded Geology Based on BP Neural Network

The interbedded geology of sandstone and mudstone is common in roadbed engineering in loess area.Due to the characteristics of soft-hard interbedded strata,it is difficult to design the blasting parameters and hard to achieve ideal blasting effect in engineering practice.A blasting parameter optimization method based on BP neural network was proposed in the paper.In this method,the main factors affecting the blasting effect were taken as the input parameters of the network,the blasting effect as the output parameters,and the collected data samples as the training and testing samples to establish the BP neural network.On the basis of the parameters of blasting scheme calculated by theory,16 kinds of blasting schemes were selected by orthogonal test method.Based on the well-trained neural network to predict the blasting effect,5 groups of schemes with better blasting effects were selected for blasting test,and the best blasting scheme was determined according to the blasting effect.The method has been applied in the excavation of cutting engineering of Hengshan Station,and good results have been obtained.