进化神经网络在定量预测回采巷道围岩移近率中的应用  

Application of Evolutionary Neural Network in the Quantitative Prediction of Surrounding Rock Displacement Rate

作　　者：王民华[1] 张召千[1] 牛显[1] 

机构地区：[1]太原理工大学矿业工程学院,山西太原030024

出　　处：《山西煤炭》2012年第5期59-61,共3页Shanxi Coal

摘　　要：神经网络处理非线性关系有明显优势。把影响回采巷道移近率的四个指标作为神经网络的输入层,巷道顶底板移近率和两帮做为网络的输出层,运用大量的回采巷道样本数据,对巷道围岩移近率的神经网络模型进行学习训练,得到稳定的网络结构;结合工程实践对训练得到的稳定网络结构进行检验,表明进化神经网络在回采巷道移近率的定量预测中有较大的实用价值。Evolutionary neural network(ENN) has obvious advantages in dealing with non-linear relation.ENN input layer consists of four indicators which influences the surrounding rock displacement rate,and ENN output layer is made up of the displacement rate of roof-floor and two sides.Huge sample data are used to train the ENN and to achieve stable network structure.The engineering practice is used to test the structure.The result shows that the ENN is practical in the prediction of displacement rate.

关 键 词：回采巷道 围岩移近率 进化神经网络 

分 类 号：TP1[自动化与计算机技术—控制理论与控制工程]