基于灰色理论与BP神经网络的特长瓦斯隧道爆破参数优选

Optimal Prediction of Blasting Parameters for Extra-long Gas Tunnel Based on Grey Theory and BP Neural Network

围岩岩体力学行为复杂多变,为避免爆破振动激发煤与瓦斯突出,获得合理的松动爆破,对特长瓦斯特大断面隧道爆破参数进行优选预测研究。基于灰色系统理论和BP神经网络,考虑围岩累积爆破损伤变形具有的动态性、瓦斯等灰色信息,选取最小抵抗线、炮孔间距、装药集中度等参数作为主要优选指标,建立基于单位化约束条件的综合集成优选模型,并对爆破振动效应下的特长瓦斯特大断面隧道的爆破参数进行优选预测。结果表明,建立的基于单位化约束条件的综合集成优选模型降低了试验中爆破参数的离散程度,当爆破参数E、W、q1的优选值分别为60、70 cm、0.12 kg/m时,基于单位化约束条件的综合集成优选模型优选预测值精度较高,隧道爆破效果较好。

The surrounding rock mass of extra-long gas tunnel is complicated and variable in mechanics behavior. In order to obtain reasonable loose blasting and avoid blasting vibration from motivating coal and gas outburst, the blasting parameters are studied for optimal prediction with respect to extra-long gas tunnel with large cross-section. Based on the grey theory and BP neural network and in consideration of the grey information of cumulative damage to the dynamic characteristics of extra-long tunnel on account of rock mass blasting and the gas, an integration optimal model based on the constraints of unitization for blasting parameters is established by means of selecting such main optimum indexes as the line of least resistance, the borehole spacing and the charge concentration, so as to fulfill optimal prediction of blasting parameters. The results show that the established integration optimal model reduces the discrete degree of test blasting parameters, when blasting parameters of E, W, q1 are 60, 70 cm, 0. 12 kg/m respectively, the accuracy of the integration optimal model is very high with good blasting effect.