【摘要】 以某露天矿为示范区,对训练样本中的自变量分别加减某个比率,构成两个新的训练样本,利用新训练样本和已训练好的神经网络,获取模拟仿真结果.计算两个训练样本模拟仿真值的差及其平均值,该平均值即为各输入参数的MIV值;以MIV值为评价依据,研究了各输入参数对最远抛掷距离、抛掷率、松散系数的影响权重;对比分析了不同MIV调节率下,各输入参数影响的权重变化情况,并应用2009-2010年实际爆破观测数据进行了相关实验.结果表明:台阶高度对最远抛掷距离、抛掷率、松散系数影响最大,其权重值在0.3～0.4之间;台阶高度与抛掷率呈正相关,与最远抛掷距离、松散系数呈负相关;平均炸药单耗对抛掷率的影响较为显著,其权重值为0.2左右,炸药单耗与抛掷率呈正相关.更多还原

【Abstract】 This research takes an open-pit mine as a case in point,and independent variable of trained sample plus or minus a ratio constitute two new training samples.Using new training sample and the trained neurtal network is to obtain the simulation results,and calculate the difference of simulation values of the two trained samples and their average values.Those average values are MIV values of input references.The weights of input reference variables effecting on max cast distance,cast blasting ratio,coefficient of volumetric expansion are studied according to MIV values.The changing trends of input reference variables′ weights are investigated under the condition of comparative analysis of different MIV ratios.The experiment is done by using of cast blasting surveying data of some open pit during 2009-2010.The results indicate: bench height is the most important factor to affect on the prediction values of max cast distance,cast blasting ratio,coefficient of volumetric expansion.Powers of bench height ranges from 0.3 to 0.4.Bench height is positive correlation with cast blasting ratio,and is negative correlation with max cast distance and coefficient of volumetric expansion.Average explosive charge is more important factor affecting on the result of cast blasting ratio,and weight of average explosive charge is about 0.2.Average explosive charge is positive correlation with cast blasting ratio.更多还原