基坑变形的神经网络预测与有限元分析-建筑与土木工程专业论文.docxVIP

摘要摘要 摘要 摘要 随着现代化建设的发展和城市化进程的加快，地上空间的开发利用已日益饱 和，愈益要求开发城市三维空间。各类用途的地下空间已经在世界各大城市中得 到开发利用，产生了大量的深基坑工程，由此引起的环境土工问题也十分突出。 而基坑又往往处于房屋和生命线工程的密集地区，如何在保证工程自身稳定的同 时，有效地控制基坑周围土体移动以保证周围环境的安全，已成为岩土工作者共 同关注的问题。 本文首先对基坑工程的发展进行了回顾，简要介绍了基坑工程中现有的变形 计算方法。针对今年来兴起的神经网络等软科学，本文介绍了神经网络的原理， 并介绍了如何利用由工程监测获得的数据形成的时间序列，用BP神经网络进行 支护墙体水平位移的预测。结合浦东南路下立交工程，利用训练好的BP网络进 行预测并和实测值进行对比，取得了较好的一致，验证了此方法的可行性。 本文利用有限单元法对基坑工程的开挖过程进行了数值模拟。对浦东南路下 立交工程选一截面进行了二维有限元的分析计算，模拟结果基本符合基坑开挖过 程的实际情况。着重将计算结果中的支护墙体的水平位移与实测结果进行了比 较，并对结果进行了分析。 本文利用BP神经网络预测和有限元分析基坑，对今后基坑工程的设计和施 工拓宽了思路。 关键词：基坑；水平位移；有限单元法；BP神经网络；Ansys AbS：tractABSTRACT AbS：tract ABSTRACT As a result of development of modemization and civilization，the exploitation and utilization of overground space become more and more difficult．So people focus on the exploitation of the underground space，gradually．All kinds of underground projects spring out in many megapolis in the world and lots of excavation projects take place．Meanwhile，many geo—environmental problems come into being．As a result of most of excavation projects locate at the place where buildings and lifelines are dense，it is very important to not only guarantee the stability of projects but make the displacements of the surrounding soil under control and assure the safety of the surroundings for geotechnicai engineers． Firstly,this paper nlake a review of the development of excavation and introduce several analyses methods of excavation berifly,Secondly,herein the spring up of“soft science”such aS neural network method，paper introduce the principle of the neural network method and how to utilize the BP neural network depict the horizontal displacements of the retaining structure．Furthermore，paper utilizes the BP neural network method to design a program to depict the horizontal displacements ofthe retaining structure．In the South Pudong road underground intersection projects， it is proved that the method is feaSible。 Finally,paper analyses the coarse of the excavation in finite element method in Ansys prog