基于BP神经网络的中深孔爆破块度预测

Prediction of blasting block of medium-depth hole based on BP Neural Network

利用神经网络极强的非线性动态处理能力,对开磷集团用沙坝矿中深孔爆破块度进行预测。选取岩体特性、炸药性能、爆破参数作为神经网络的输入层,平均块度和大块率作为输出层,建立神经网络预测模型,进行块度预测。对比神经网络预测和工业试验所得的数据,表明神经网络预测爆破块度能取得良好的效果。该方法对于优化中深孔爆破参数具有重要意义。

The neural network,with porcessing cability of a strong nonlinear dynamic problem,was adopted to predict blasting block of medium-depth hole in Yongshaba Mine.In the analysis,the paper selects the rock mechanics properties,explosive performance,blasting parameters as input layer of neural network,the average degree and large block rate as the output layer to establish the neural network model,which shows that the prediction of blasting block with neural networks can achieve a better result by comparing the data obtained by neural network and industrial experiment.This method has significance in parameters optimization of medium-depth hole blasting.