摘要
为提升大跨度铁路连续刚构-拱桥有限元模型预测精度,同时针对传统有限元模型修正中只采用全局最优解而忽略可能更接近实际情况的局部最优解问题,提出一种基于距离机制的改进稳态遗传算法(DSSGA),利用实测静动力数据对其初始有限元模型进行模型修正。首先,介绍桥梁基本信息、初始有限元模型和静动力试验及其试验结果;其次,介绍DSSGA算法的理论以及其结合Kriging代理模型的基本修正流程,并通过测试函数验证DSS-GA算法的优化效果;最后,通过灵敏度分析选择待修正结构参数,利用拉丁超立方设计构建Kriging代理模型并检验其精度,利用静力位移、试验模态参数构造目标函数,对该桥进行模型修正。结果表明:与标准稳态遗传算法(SSGA)相比,DSSGA算法能够提供目标函数在搜索域的全局最优解和更多组局部最优解,有效避免SSGA算法角度机制所产生的解集不完整的局限性,且全局最优解的目标函数值更小,具有更高的搜索效率。经过模型修正,所有测点的位移相对误差控制在10%以内,频率相对误差控制在5%以内,修正后模型的预测精度大幅度提升。修正后的模型可作为该桥的基准有限元模型,用于后续桥梁健康监测与状态评估。
This study proposes a distance-mechanism-based improved steady-state genetic algorithm(DSSGA)to increase the prediction accuracy of the finite element(FE)model of the long-span railway continuous rigid frame-arch bridge and to target the problem of using only the global optimal solution in the traditional FE model updating but ignoring the multiple local solutions which may be closer to the actual situation.The initial FE model of the bridge is updated by experimental measured static and dynamic responses.Firstly,the fundamental information,the initial FE model,the in-situ static and dynamic tests,and its results of the bridge are introduced.Then,the theory of DSSGA algorithm and basic updating process combined with the Kriging surrogate model are introduced.The DSSGAs effectiveness is verified by test functions.Finally,the sensitivity analysis is used to select the structural parameters to be updated,the Kriging surrogate model is established through the Latin hypercube,then its accuracy is checked.The bridge is updated with an objective function established based on static displacement and experimental modal parameters.The results indicate that compared with the standard steady-state genetic algorithm(SSGA),the DSSGA provides the global and more local solutions of the objective function in the search domain,effectively avoiding the limitation of incomplete solution sets generated by the angle mechanism of SSGA.Additionally,the global solution obtained from DSSGA is smaller than that obtained from SSGA.After model updating,the displacement and frequency relative errors are controlled under 10%and 5%respectively.The updated FE models prediction accuracy is significantly improved,which can serve as the bridges baseline FE model for subsequent structural health monitoring and condition assessment.
作者
梅冲
宋任贤
周云飞
霍学晋
秦世强
MEI Chong;SONG Renxian;ZHOU Yunfei;HUO Xuejin;QIN Shiqiang(Wuhan Light Industry Engineering Technology Co.,Ltd.,Wuhan 430040,China;School of Civil Engineering and Architecture,Wuhan University of Technology,Wuhan 430074,China;China Railway Major Bridge Reconnaissance and Design Institute Co.,Ltd.,Wuhan 430074,China)
出处
《铁道标准设计》
北大核心
2024年第7期108-117,共10页
Railway Standard Design
基金
国家自然科学基金项目(51608408)
中央高校基本科研业务费专项资金资助项目(2017-IVB-046)。
关键词
桥梁工程
改进稳态遗传算法
模型修正
多解问题
连续刚构-拱组合体系
目标函数
代理模型
bridge engineering
steady-state genetic algorithm
model updating
multisolution problem
continuous rigid frame-arch railway bridge
objective function
surrogate model
