摘要
由于加载系统的非线性、不确定性以及振动台-试件耦合特性,非迭代加载的振动台试验的复现精度较差;而迭代加载过程常对试件造成损伤,进而改变试件力学性能。针对此问题,提出一种基于替代试件模型的振动台迭代控制方法。该方法首先构建与试件动力特性相近的替代模型,再利用该模型完成迭代学习得到振动台加载命令,最后开展原试件的振动台试验。本文重点研究了获取替代试件模型动力参数的方法。研究表明:针对线性试件,基于试件模态参数的反解法以及基于遗传算法的优化方法均能获得替代效果良好的线性模型参数;对于非线性试件,考虑阻尼器参数的线性替代模型能实现更好的替代效果。
Due to the nonlinearity and uncertainty of loading systems and coupling characteristics between shaking tables and specimens,acceleration reproducibility in shaking table tests without iterative loading is often unsatisfactory.The iterative loading process often causes damage to the specimen,and then changes the mechanical properties of the specimen.To resolve these issues,an iterative control method based on a surrogate specimen model for shaking table tests was proposed.For this method,a surrogate specimen model possessing similar dynamic characteristics to the original specimen need to be firstly designed and built.After completing iterative learning to obtain shaking table commands with this surrogate model,shaking table tests for the original specimen can be conducted.This paper focuses on methods of obtaining dynamic parameters of surrogate specimen models,namely two methods based on specimen modal parameters and genetic algorithm optimization,respectively.Results show that both methods can provide good replacement effects for linear original specimens.For nonlinear specimens,the linear surrogate model considering a damper resulted from the genetic algorithm optimization can accurately reproduce original shear force time histories.
作者
王贞
陆文龙
吴斌
贾学军
Wang Zhen;Lu Wenlong;Wu Bin;Jia Xuejun(School of Civil Engineering and Architecture,Wuhan University of Technology,Wuhan 430070,China;Hainan Institute of Wuhan University of Technology,Sanya 572000,China;Nanjing Technology University,Nanjing 211816,China;China Construction Second Engineering Bureau Ltd,Wuhan 430070,China)
出处
《土木工程学报》
EI
CSCD
北大核心
2022年第S01期203-208,256,共7页
China Civil Engineering Journal
基金
国家自然科学基金(52078398)
海南省重大科技计划项目(ZDKJ2021024)
武汉理工大学新教师科研启动基金(2021IVA031)
武汉理工大学三亚科教创新园开放基金(2020KF0011)
关键词
振动台
替代模型
试验方法
遗传算法
迭代学习
shaking table
surrogate model
test method
genetic algorithm
iterative learning
作者简介
王贞(1983—),男,博士,特任研究员。主要从事结构抗震试验方法研究;陆文龙(1998—),男,硕士研究生。主要从事结构混合试验方法研究;吴斌(1970—),男,教授,博士。主要从事结构抗震试验方法和结构振动控制研究;贾学军(1978—),男,博士研究生,高级工程师。主要从事建筑施工、智慧建造等方面的研究。
