SOM神经网络预测方法在基坑变形中的运用  

Application of SOM Neural Network in the Foundation Pit of Tunnel

作　　者：丁杨[1] 陈希杰 杜欣 胡洋[3] DING Yang CHEN Xi - jie DU Xin HU Yang(Department of Civil and Architecture, East China Jiaotong University, Nanchang Jiangxi 330013, China Shanghai Foundation Engineering Group Co. , Ltd, Shanghai 200002, China School of architecture and engineering, Nanchang Institute of Technology, Nanchang Jiangxi 330099, China)

机构地区：[1]华东交通大学土木建筑学院,江西南昌330013 [2]上海市基础工程集团有限公司,上海200002 [3]南昌工程学院建筑工程学院,江西南昌330099

出　　处：《安徽理工大学学报（自然科学版）》2017年第4期32-35,共4页Journal of Anhui University of Science and Technology:Natural Science

基　　金：国家自然科学基金资助项目(51168015)

摘　　要：城市地铁建设正逐步进入快速有序的发展阶段,各种类型的地铁事故也时常发生。因此,在隧道基坑工程中需要一种预测方法来合理的避免事故发生。针对上海市轨道交通17号线上的某车站站所产生的深层水平位移问题,运用MATLAB神经网络工具箱仿真并建立SOM神经网络预测模型。实验结果表明,通过输入已知数据建立的SOM神经网络预测变形曲线与实测位移的绝对误差值在0.123~1.43mm之间,误差值范围小,在实际工程中是可以接受的。因此,建立SOM神经网络模型对于基坑变形问题有很好的预测能力,该方法为地下工程提供了新的预测手段。The construction of urban subway is entering the stage of rapid and orderly development, but it was reported subway accidents of different types occurred frequently. Therefore, it is necessary to adopt a forecasting method to avoid accidents. In order to solve the deep horizontal displacement problem in Shanghai rail traffic No 17 line station Cao Ying Lu, neural network toolbox of MATLAB simulation was used and SOM neural network prediction model was established. The experimental results showed that SOM neural network was established by the input data to predict the absolute error of deformation curve and the measured displacement values between 0. 123 -1.43mm, the error range was small, which is acceptable in the actual project. Therefore, the SOM neural network model has a good ability to predict the deformation of the foundation pit.

关 键 词：隧道工程 车站 SOM神经网络 MATLAB分析 

分 类 号：U455[建筑科学—桥梁与隧道工程]