欢迎光临《计算力学学报》官方网站！

System reliability assessment of long-span cable-stayed bridges using an updating support vector algorithm

DOI：10.7511/jslx201502003

 作者 单位 E-mail 刘扬 长沙理工大学 桥梁结构安全控制湖南省工程实验室, 长沙 410114 鲁乃唯 长沙理工大学 桥梁结构安全控制湖南省工程实验室, 长沙 410114 lunaiweide@163.com 殷新锋 长沙理工大学 桥梁结构安全控制湖南省工程实验室, 长沙 410114

针对斜拉桥静力体系可靠度分析中隐式功能函数重构和繁杂失效路径的特点,提出了一种基于更新支持向量的体系可靠度分析方法,将传统的用于构件可靠度分析的支持向量机(SVM)改进并应用于斜拉桥体系可靠度分析。该方法主要有4个步骤:首先通过构件的敏感分析识别斜拉桥的主要失效路径;其次采用最小二乘支持向量机(LS-SVM)对斜拉桥的隐式功能函数进行重构,并通过Monte-Carlo抽样得出构件的可靠指标;然后根据已经更新的有限元模型对支持向量进行更新,得出相关构件失效后的剩余构件的条件可靠指标;最后由结构体系的失效树和串并联关系得出斜拉桥的体系可靠度。主跨为420 m的混凝土斜拉桥算例分析表明了上述算法的有效性和实用性,同时也获得了该斜拉桥的主要失效路径并识别了影响其体系可靠度的主要构件。

A new system reliability assessment method is proposed based on updating support vector method, considering the characteristic of cable-stayed bridges with explicit performance functions and multiple failure modes.The traditional Support Vector Machine (SVM) is developed and applied to system reliability analysis of cable-stayed bridges.There are four main steps in this algorithm:firstly, the main failure sequences should be rectified according to the sensitivity analysis of components;secondly, Least Squares Support Vector Machine (LS-SVM) is used to approximate the structural responses, and then reliability indices of components are obtained using Monte-Carlo simulation method.Thirdly, the support vectors in LS-SVM model are updated on the basis of updated finite element models, and then the conditional reliability indices of the rest components is calculated based on the updated support vector model;finally, the failure tree of structural system and series parallel method are used to estimate the system reliability index.The numerical analysis of a concrete cable-stayed bridge with 420m span has shown the effectiveness and applicability of the proposed algorithm.At the same time, the main failure sequences and the main failure components of the concrete bridge are given.