基于GRA-BA-RBFNN模型的露天矿台阶爆破岩石位移预测

Prediction of Rock Displacement for Step Blasting in Open-pit Mine Based on GRA-BA-RBFNN Model

露天矿爆破开采产生的岩体位移将影响矿石的分布,从而造成矿石的贫化或损失。为了较准确地把握爆破后矿石分布情况,采用径向基函数神经网络对爆破后岩石的位移进行预测。由于爆破后岩石的位移十分散乱,无法将其具体到爆破的每个部位,故将其转化为剖面多边形的质量中心进行整体考虑,由此完成爆破后岩体位移的量化过程。同时,针对样本数量不足的情况,引入GRA理论,确定影响爆破后岩体位移的主要因素;利用RBFNN函数预测爆破后岩石位移的适应能力和稳定性,并采用BA算对RBFNN函数的径向基扩展速度进行确定,从而建立GRA-BA-RBFNN预测模型。最后,使用该模型对江西省德兴铜矿爆破后爆堆的质心位移进行了预测,对比未提取主要因素时的RBFNN模型和未经BA算法优化的RBFNN模型的预测结果,发现模型的精度和稳定性都有了很大的提高,该研究可以为露天矿爆破的岩石位移预测提供一定的借鉴意义。

The rock displacement produced by blasting mining in open-pit mines would affect the ore distribution and result in ore dilution or loss.In order to accurately obtain the ore distribution after blasting,the radial basis function neural network(RBFNN)was used to predict the rock displacement after blasting.Because the rock displacement after blasting was very scattered,it cannot be specific to every part of blasting.Therefore,it was transformed into the mass center of section polygon for overall consideration,so as to complete the quantification process of rock displacement after blasting.At the same time,in view of insufficient samples,GRA theory was introduced to determine the main factors affecting rock displacement after blasting.RBFNN function was used to predict the adaptability and stability of rock displacement after blasting,and BA calculation was used to determine the radial basis expansion speed of RBFNN function.Thus,the GRA-BA-RBFNN prediction model was established.Finally,the centroid displacement of blasting pile in Dexing Copper Mine in Jiangxi Province was predicted by this model.Compared with the predicted results of RBFNN model without extracting the main factors and RBFNN model without BA algorithm optimization,it was found that the accuracy and stability of the model was greatly improved.This study can be used as a reference in rock displacement prediction of open-pit mine blasting.