基于人工神经网络的中深孔爆破参数优选

Optimization Selection of Medium-deep Hole Blasting Parameters Based on Artificial Neural Network

为了得到合理的中深孔爆破参数,以矿岩的弹性模量、容重、抗拉强度、抗压强度、摩擦角以及黏结力作为输入因子,以炮孔的崩矿步距、孔底距以及炸药单耗为输出因子,以国内爆破工艺相类似矿山的相关数据为训练样本,建立BP神经网络模型进行爆破参数优选。以港里铁矿为工程背景,通过优选和影响因素分析,得到崩矿步距1.67 m,孔底距1.8 m,炸药单耗0.43 kg/t的爆破参数。使用后证明,比原炸药单耗(0.52 kg/t)降低了17.3%。

In order to obtain the reasonable medium-deep blasting parameters,taking the ore-bearing rock elastic modulus,bulk density,tensile strength,compressive strength,friction angle and cohesive force as the input factors,taking the hole collapse interval,depth of holes,and consumption of dynamite as the output factors,taking the relative datas of the mines that the blasting technology are similar with each other as the training samples so as to establish the BP neural network model to conduct blasting parameters optimization. Taking the Gangli iron mine as the engineering background,through optimization and analysis of the influence factors to obtain the caving step distance is 1. 67 m,the hole bottom distance is 1. 8 m,explosive consumption is 0. 43 kg / t. The application results show that,the optimized explosive consumption is reduced by 17. 3% relative to the original unit explosive consumption（ 0. 52 kg / t）.