地铁盾构施工地表变形的神经网络预测及应用

Prediction and application of neural network on surface deformation from subway shield tunneling construction

为对城市地铁盾构隧道掘进引发的地表变形进行预测,以太原地铁某区间盾构隧道工程为例,通过敏感性分析方法确定影响盾构施工地表变形的主要土层参数与施工参数。在此基础上以主要影响因素为设计因子开展正交试验,将试验结果作为训练样本及测试样本,建立径向基函数(radial basis function,RBF)神经网络模型用于盾构施工地表变形的预测。结果表明,建立的RBF神经网络模型具有较高的精度,能够较好地反映地表变形与各主要影响因素之间的非线性关系,可以用于盾构施工地表变形的预测。

In order to predict the surface deformation caused by shield tunneling in urban subway,the tunnel shield tunnel project in Taiyuan was used as an example to determine the main soil parameters and construction parameters that affect the surface deformation through the sensitivity analysis method.Orthogonal experiments were conducted with the main influencing factors as design factors.The test results were used as training samples and test samples to develop radial basis function (RBF) neural network models for prediction of surface deformation in shield construction.The results show that the established RBF neural network model has higher precision and can better reflect the nonlinear relationship between surface deformation and main influencing factors.It can also be used to predict the surface deformation from shield construction.