岩壁梁爆破参数优化的神经网络模型

NEURAL NETWORK MODEL OF BLASTING PARAMETERS OPTIMIZATION OF CRANE BEAM AT ROCKWALL

地下厂房岩壁梁开挖后不仅要求有光滑的轮廓面,而且破坏范围最小,因此开挖难度大、质量要求高。在开挖该部位时常通过工程类比和在相似地段进行现场爆破试验,来决定爆破开挖参数,这种开挖方法成本高,有时效果也不理想。本文选择有代表性的实际爆破参数为样本,确定炸药类型、岩体裂隙发育程度、孔径、孔深、线装药密度、最小抵抗线、孔间距为影响爆破开挖效果的主要因素,应用神经网络建立岩壁梁爆破参数优化模型,进行爆破参数优化设计,并与爆破试验得出试验参数相比较,结果表明设计值与现场试验值吻合较好。对爆破试验部分进行声波检测也表明其爆破质量最好,其爆破松动圈范围最小。

Being required to leave less damage and to produce a smooth outline face, the high quality of excavation of crane beam at rockwall is specified in the excavation,and as a result, the excavation is a very difficulty one. Commonly the excavation parameters are determined by consulting the similar engineering and doing site blasting tests. But this method is subject to high operation cost and not ideal blasting effect. In this present paper, the optimum model for blasting parameters of anchoring rock beam is established by making use of a strong mapped function of the neural network technology with the typical samples of other practical smooth blasting parameters,like explosive type,crack of rock body and developing degree, hole diameter, hole depth, line charging density and minimum burden, as the main factors influencing the excavation effect. The experimental blasting parameters for crane rock beam are determined with the model, and are compared with the ones obtained from the experiments in-slte with similar conditions. The results show that the in-slte experimental blasting parameters are preferably identical to those of the optimum design. The examination of the blasting acoustic wave of the protective layer and platform of rock mass indicates that the blasting, effect is satisfied and the loose ring of surrounding rock mass is the smallest.