基于神经网络的露天矿爆破参数优化研究

Optimization of the Open-pit Mine Blasting Parameters Based on Neural Network

马家塔露天矿爆破效果不理想,大块率较高、炸药单耗偏高、残留根底现象严重。为解决这些问题,基于BP神经网络建立了露天矿爆破参数优化模型。采用该矿的实际爆破参数作为样本数据对模型进行了训练,通过实例数据的检验证明该模型具有较高的预测精度。经过现场试验和模型仿真分析,采用优化后的爆破参数进行了爆破,取得了较好的效果:大块率控制在2%以下,炸药单耗由0.41 kg/m3降为0.37 kg/m3,每延米炮孔爆破量由17 m3/m增加至23 m3/m。

The blasting effect is not ideal in Majiata open-pit mine, such as higher rate of blocky ore, high explosives consumption, and abundant residuals. To solve these problems, an optimization model of the open-pit mine blasting parameters based on BP neural network is built. The actual blasting parameters are applied into the model as the sample data. The tests with practical data show that the model has higher prediction accuracy. With the on-site tests and model simulation analysis, a good blasting result is obtained by using the optimized parameters. Large block rate is controlled below 2%, explosives consumption reduced from 0.41 kg/m3 to 0.37 kg/m3 , blasting volume of each hole increased from 17 m3/m to 23 m3/m.