论文部分内容阅读
为控制隧道拱顶下沉过程及其最终下沉量,保障隧道工程建设安全,需要准确预测隧道拱顶沉降量。基于有限元和人工神经网络,结合工程实践,提出ANSYS有限元和BP神经网络相结合(ANSYS-BP)的隧道拱顶沉降预测方法。其预测步骤为:根据隧道地质条件、围岩性质、施工工艺、支护衬砌方式,采用ANSYS建立隧道模型,计算多步工艺施工下隧道拱顶指定单元处沉降值序列,以此构建BP神经网络预测模型,将沉降序列信息作为输入样本对BP神经网络的进行训练、测试。并以某在建公路隧道为研究对象,进行预测分析。结果表明,ANSYS-BP方法能够有效改善隧道拱顶沉降的预测性能,其预测值与实际监测值基本一致,预测精度高于单独使用ANSYS计算或BP网络预测的结果。
To control the process of tunnel dome doming and its final settlement, to ensure the safety of tunnel construction, it is necessary to accurately predict the settlement of tunnel vault. Based on the finite element method and the artificial neural network, combined with the engineering practice, a prediction method of tunnel vault settlement based on the combination of ANSYS finite element and BP neural network (ANSYS-BP) is proposed. The prediction procedure is as follows: According to the tunnel geological conditions, the nature of the surrounding rock, the construction technology and the supporting lining method, the tunnel model is established by ANSYS, and the settlement value sequence of the designated unit of the tunnel vault under the multi-step process is calculated to construct the BP neural network Prediction model, the settlement sequence information as input samples of BP neural network training and testing. And a certain road tunnel under construction as the research object, the prediction analysis. The results show that the ANSYS-BP method can effectively improve the prediction performance of the tunnel vault settlement. The predicted value is basically consistent with the actual monitoring value, and the prediction accuracy is higher than that of ANSYS alone or BP network prediction.