基于混合算法的大跨度斜拉桥可靠度评估  

Reliability Assessment of Long-span Cable-stayed Bridges Based on Hybrid Algorithm

作　　者：刘扬[1,2] 鲁乃唯[1,2] 汪勤用 

机构地区：[1]长沙理工大学土木与建筑学院,湖南长沙410114 [2]桥梁工程湖南省高校重点试验室,湖南长沙410114

出　　处：《公路交通科技》2014年第7期72-79,共8页Journal of Highway and Transportation Research and Development

基　　金：国家自然科学基金项目(51378081;51108045);教育部新世纪优秀人才支持计划项目(NCET-10-0139);湖南省研究生科研创新项目(CX2013B366)

摘　　要：为研究大跨度预应力混凝土(PC)斜拉桥的可靠度评估问题,提出了适用于大跨度PC斜拉桥这类复杂结构可靠性分析的混合算法。该方法综合运用了有限元分析(FEA)、径向基函数(RBF)神经网络、遗传算法(GA)和Monte Carlo重要抽样(MCIS)方法,并对算法中的关键步骤(RBF神经网络的初始样本点设计方法、MCIS的抽样中心点位置等)进行了改进,使结构分析模块与可靠度计算模块智能结合。利用数值算例的可靠度分析对该算法的有效性进行了验证。最后,以一座主跨为420 m的双塔PC斜拉桥为工程背景,进行了正常使用极限状态下的可靠度分析。参数分析表明:在汽车荷载作用下,该斜拉桥的主梁跨中位移超限失效概率比最长斜拉索强度失效概率高;汽车荷载的均值和标准差是影响斜拉桥可靠度的重要因素;随着汽车荷载均值系数的增大,主梁跨中位移超限失效的可靠指标下降的趋势较为显著。In order to analyze the reliability assessment of long-span prestressed concrete （PC） cable-stayed bridge, a hybrid algorithm which applies to complex structures such as long-span PC cable-stayed bridges is proposed. The method combined finite element analysis （FEA）, radial basis function （RBF） neural network, genetic algorithm （GA） and Monte-Carlo importance sampling method （MCIS）, and modified some critical steps like initial sample point design method of RBF neural network, MCIS sampling center position in the algorithm to intelligently combine the structural analysis module and the reliability calculation module. The effectiveness of this algorithm is verified by reliability analysis of a numerical example. Finally, reliability analysis under serviceability limit state is processed with the background of a 420 m main-span 2- pylon PC cable-stayed bridge. The parameter analysis indicates that （1） under vehicular load, the failure probability of girder mid-span displacement transfinite is higher than that of the longest cable strength; （2） the mean value and standard deviation of vehicular load exhibit higher influence on reliability of the cable- stayed bridge; （3） the reliability index of transfinite failure of girder mid-span displacement has a distinct decrease tendency with the increase of mean value coefficient of vehicular load.

关 键 词：桥梁工程 斜拉桥 混合算法 神经网络 可靠指标 安全评估 参数敏感性 

分 类 号：U448.27[建筑科学—桥梁与隧道工程]