基于进化径向基神经网络方法的爆破参数优化

Blasting Parameters Optimization Based on Evolutionary Radial Basis Function Neural Network

多宝山铜矿处于高寒地区,常年的冻融循环作用以及台阶爆破的扰动影响了矿岩台阶的物理力学性质,按照原有的爆破参数进行设计并施工,爆破效果往往不理想,主要是大块率偏高、底根较多。为此,采用进化径向基神经网络方法对孔网参数、炸药单耗、排间延期时间等爆破参数进行优化,根据神经网络的训练与预测结果,最终得出了一套适用于多铜矿岩台阶爆破的最优爆破参数,通过进行一系列的现场爆破试验,并与之前的爆破效果进行对比,大块率显著降低,炸药单耗有所下降,提高了铲装的工作效率,节省了爆破成本,取得了良好的爆破效果,并增加了采矿经济效益。此方法科学可行,适用于多宝山铜矿的台阶爆破参数优化。

Duobaoshan Copper Mine is located in the alpine region,the perennial freeze-thaw cycle and the disturbance of step blasting affect the physical and mechanical properties of the step.According to the original blasting parameters for design and construction,the blasting effect is often not ideal,the main performance is the bulk ratio is higher and the blasting bottom is more than expected goal.The radial basis function neural network is adopted to optimize the parameters of hole mesh,single consumption of explosive,delay time between row and other blasting parameters.According to the training and prediction results of neural network,a set of optimal blasting parameters for multi-copper ore step blasting is obtained.By conducting a series of field blasting tests and comparing them with the previous blasting results,the results show that the efficiency of shoveling is improved,the blasting cost is saved,the blasting effect is good and the economic benefit of mining is increased.The method is scientific,feasible and suitable for optimization of step blasting parameters of Duobaoshan Copper Mine.