基于RBF神经网络的爆破参数优选试验研究

发布时间：2018-08-09 12:14

【摘要】：为了优化某矿山采场的爆破参数,提出了水平凿岩方式代替上向倾斜凿岩方式的改进方案。根据改进方案,在采场中进行了L_9(3~3)的爆破参数正交试验,获取了9组试验结果。利用RBF神经网络模型对试验结果进行预测,以最小抵抗线、孔间距、周边孔距作为输入因子,炸药单耗、大块率作为输出因子,当隐含层个数为9时,预测效果最优;在综合考虑爆破成本和爆破效果的前提下,提出了爆破综合期望指数公式来最终优选爆破参数,获取了48个爆破综合期望指数,最大值1.134。综合分析,最终推荐矿山最优爆破参数为:排距1 m,孔间距1.4 m,周边孔距1 m,炸药单耗0.185 kg·t~(-1)。实际应用证明,选择的孔网参数合理,大块率降低至7%以下,极大的降低了爆破成本。
[Abstract]:In order to optimize the blasting parameters of a mine stope, an improved scheme of horizontal rock drilling instead of upward inclined rock drilling is put forward. According to the improved scheme, the orthogonal test of blasting parameters of L9 (3K3) was carried out in the stope, and nine groups of test results were obtained. The RBF neural network model is used to predict the test results. The minimum resistance line, the distance between the holes and the distance between the peripheral holes are taken as input factors, the unit consumption of explosives and the bulk rate are taken as the output factors. When the number of hidden layers is 9, the prediction effect is the best. On the premise of synthetically considering blasting cost and blasting effect, this paper puts forward the formula of blasting comprehensive expectation index to select blasting parameters, and obtains 48 blasting comprehensive expected exponents, the maximum value of which is 1.134. Through comprehensive analysis, the optimum blasting parameters are recommended as follows: row distance 1 m, hole spacing 1.4 m, peripheral hole spacing 1 m, explosive consumption 0.185 kg t ~ (-1). The practical application shows that the parameters of the selected hole mesh are reasonable and the block ratio is reduced to less than 7%, which greatly reduces the blasting cost.
【作者单位】：中南大学资源与安全工程学院;
【基金】：国家科技支撑计划项目(2013BAB02B05)
【分类号】：TD235

【相似文献】