近年来隧道方面的国家自然科学基金项目摘要.docVIP

隧道方面的自然基金项目摘要 1批准目名称隧道工程信息化设计与智能分析方法研究项目类别面上项目申请代码E0807 项目负责人乔春生负责人技术职称教授依托单位北京交通大学资助年限01/01/2001 - 01/12/2003 资助经费17(万元)中文摘要：以新奥法为基础的隧道工程信息化设计思想已提出了几十年，但目前我国隧道设计仍然以工程类比法为主。这主要是由于信息化设计的各个环节严重脱节，缺乏功能齐全、使用方便、形象直观的计算与分析工具。针对这一问题，本项目通过理论分析和试验研究，重点进行了隧道工程三维可视化快速仿真方法研究及计算机程序开发、隧道工程监控量测数据智能处理方法和隧道变形预测的智能方法研究。对有限元求解器进行了改进，使计算速度和效率明显提升；提出了新的地下工程可视化三维建模方法和计算结果的三维可视化方法，以此为基础编写了岩土工程三维有限元可视化仿真程序；首次将以统计学习理论为基础的机器学习新方法- - 支持向量机回归算法引入隧道工程设计，在传统支持向量机一元回归算法基础上，提出了支持向量机多元回归算法和隧道位移反分析与位移预测的智能方法，解决了传统反分析方法需要事先假设围岩力学模型的难题，与人工神经元网络相比，可以较好地解决局部优化、样本数量不足等问题。通过对浅埋单线电气化铁路隧道的应用证明，这种新的智能分析方法计算误差小，使用方便，可满足现场使用要求。研究成果将有利于提高我国的隧道建设水平。中文主题词：隧道工程，可视化计算，支持向量机，反分析，变性预测英文摘要:The informative design iedr of tunnel based on the New Austrilian Tunnel Method has been propsed for decades, but the tunnels of our country is generally designed based on the engineering experience and the performance of existing tunnels in similar ground at present. This is mainly because of disconnect of links for NATM and lack multiple functional, easy to use, computation and analyze tools. In order to improve the actuality mentioned above, some experiments in situ and analytical studies about visual simulation method and software can beused to tunnel engineering as well as intelligent prediction method of tunnel deformation were carried out. A fast three-dimensional finite element simulation method and a visual computing program which can be used to simulate procedures of the tunnel excavation and support was developed. The complex three-dimensional shape models can be built up easily using the program. A new intelligence method for displacement back analysis and prediction of tunnel deformation as well as the design of tunnel shotcrete-bolting support parameters based on support vector machine (SVM) was proposed. It has stronger generalization ability because the SVM theory is based on the minimization of structure risk principle. The algorithm of SVM is a convex quadratic optimization problem, therefore the soluti