摘要
以兰渝铁路线纸坊隧道为工程背景,现场监控量测资料为依据,基于Matlab中的最小二乘原理预测围岩最终位移值,再分别基于BP神经网络和径向基神经网络进行位移反分析,建立了两种基于Matlab神经网络工具箱的隧道围岩位移反分析系统,对比分析了径向基神经网络相对于BP神经网络的优越性.为保证系统训练样本符合工程实际,利用Midas-GTS软件对实际隧道开挖过程(七步开挖法)进行模拟,均布构造训练样本.最后通过处理后的工程量测位移值分析反演出量测点的实际围岩参数.
Taking Zhifang Tunnel in Lan-Yu Railway as an example and the field test data as the reference,the end displacement of surrounding rock is predicted through the least-squares principle of Matlab,and the displacement back analysis is carried out based on BP neural network and Radial Basis Function（RBF）neural network,then two systems about tunnel surrounding rock displacement back analysis are established based on the Matlab neural network toolbox,comparing and analyzing the superiority of the RBF neural networks to the BP neural network.Midas-GTS is used for simulation（seven-step excavation method）in order to ensure that the systematic training samples comply with the actual project,distributing the construction training samples averagely.The parameters of the rock of actual measurement point are inversed by analyzing the actual measuring displacement at last.
出处
《兰州交通大学学报》
CAS
2010年第4期84-87,共4页
Journal of Lanzhou Jiaotong University
基金
铁道部兰渝铁路线科研项目(2009G009-B-2)
