基于神经网络的岩土工程结构随机有限元分析  

Finite Element Monte-Carlo Method Using Neural Networks for Geotechnical Reliability Analysis

作　　者：邓建[1] 朱合华[1] 

机构地区：[1]同济大学地下建筑与工程系,上海200092

出　　处：《同济大学学报（自然科学版）》2002年第3期269-272,共4页Journal of Tongji University:Natural Science

基　　金：中国博士后科学基金资助项目

摘　　要：针对有限元蒙特卡罗法计算量大的弊端和岩土工程结构功能函数不能用显式表达的可靠性分析问题 ,提出并论证了基于神经网络的随机有限元 (有限元蒙特卡罗 )分析法 .神经网络具有高度非线性的映射能力 ,可用来逼近结构响应量与随机变量的映射关系 .通过典型岩土工程结构的应用实例分析 。A finite element Monte-Carlo method using artificial neural networks(ANN) is presented in this paper.It is especially useful in such reliability analysis problems as those whose performance funtions are implicit.ANN,a universal approximator,is used to approximate the implicit mapping between structural response variables and basic random variables.ANN is trained and validated by samples generated from selected structural analyses.The tyained ANN is used to compute the response variables and then the reliability index.The new method is proved to be theoretically correct.The application and prospect of the method are illustrated by examples of geotechnical engineering structures'reliability analysis.

关 键 词：神经网络 有限元蒙特卡罗法 可靠性分行 结构分析 岩土工程 非线性映射能力 

分 类 号：TU311.2[建筑科学—结构工程] TP183[自动化与计算机技术—控制理论与控制工程]