基于神经网络的斜拉桥非线性随机静力分析  

Stochastic Static Analysis of Cable-Stayed Bridges Including Geometrical Nonlinearity Based on Artificial Neural Network

作　　者：陈铁冰[1] 谭也平[2] 

机构地区：[1]深圳职业技术学院建筑与环境工程学院,广东深圳518055 [2]深圳大学建筑与土木工程学院,广东深圳518060

出　　处：《重庆建筑大学学报》2006年第4期42-46,55,共6页Journal of Chongqing Jianzhu University

基　　金：建设部科研攻关项目(04-2-083)

摘　　要：对于大跨度柔性斜拉桥的设计和结构分析来说,考虑结构几何非线性效应进行静力分析是一项重要的内容。但是,材料、几何尺寸、外部荷载等的随机性将影响计算结果。本文基于神经网络提出了斜拉桥非线性随机静力分析方法。通过若干次确定性有限元数值抽样对神经网络训练,训练后的神经网络能够模拟随机变量与结构响应之间的非线性映射关系。算例研究表明,各种随机变量的变异对斜拉桥结构响应影响规律不同。同其他随机变量相比,斜拉索弹性模量变异对斜拉桥跨中挠度影响最显著。主梁截面面积变异对斜拉桥跨中挠度和尾索索力具有显著的影响。The static analysis of long - span and slender cable - stayed bridges including geometrical nonlinearity is an important task in design and structural analysis. However, the results from the static analysis will be variable when randomness in material properties, physical dimensions and loads have to be taken into account. In this paper, an approach of stochastic static analysis of cable - stayed bridges including geometrical nonlinearity is presented based on artificial neural network （ANN）. The ANN can be trained using a small set of numerical values obtained from the deterministic finite element method for cable - stayed bridges and the trained ANN can map the structural responses and stochastic variables such as material properties, physical dimensions and loads. A numerical example is given. The results show that the variation of the stochastic variables has a different influence to the structural responses of cable - stayed bridges. The variation of Young＇s modulus of cables has a significant influence to the deflection of cable - stayed bridges at the middle of span compared to other stochastic variables. The variation of sectional area of girders has influence to the deflection of cable - stayed bridges at the middle of span and forces of cables anchored at the end of side span.

关 键 词：冲经网络 斜拉桥 几何非线性 随机静力分析 

分 类 号：TU311[建筑科学—结构工程]