基于敏感性分析的降参模型更新混合模拟方法  

Sensitivity analysis-based model updating hybrid simulation method using reduced-parameter technique

作　　者：陈再现 刘铖 王纪伟 钟炜彭 李明刚 Chen Zaixian;Liu Cheng;Wang Jiwei;Zhong Weipeng;Li Minggang(College of Ocean Engineering,Harbin Institute of Technology(Weihai),Weihai 264209,China;Key Lab of Civil Engineering Structure and Disaster Prevention in Universities of Shandong,Harbin Institute of Technology(Weihai),Weihai 264209,China;China Construction Third Bureau First Engineering Co.,Ltd.,Wuhan 430040,China)

机构地区：[1]哈尔滨工业大学(威海)海洋工程学院,威海264209 [2]哈尔滨工业大学(威海)山东省高等学校土木工程结构与防灾实验室,威海264209 [3]中建三局第一建设工程有限责任公司,武汉430040

出　　处：《东南大学学报（自然科学版）》2024年第4期877-884,共8页Journal of Southeast University：Natural Science Edition

基　　金：国家自然科学基金资助项目(52078165)。

摘　　要：针对敏感性分析在识别参数数目过多时存在识别效率较低甚至无法分析的问题,引入分类敏感性分析和分阶段模型更新,提出基于敏感性分析的降参模型更新混合模拟方法,从而大幅降低分析和更新的参数数目,提升参数识别效率.以钢框架模型为例,选择8个模型参数,分别采用2种降参策略,讨论以不同敏感性占比作为核心参数选择标准时的降参模型更新混合模拟效果.数值模拟结果表明:不同策略下的模型更新误差差距在0.5%以内;采用分类、分阶段降参策略与全参数策略相比,参数识别的运算量减小,收敛步数大幅减少,验证了所提方法提升混合模拟参数识别速度的可行性与有效性.Since sensitivity analyses have the problem of low efficiency or even inability to analyze when dealing with a large number of identification parameters,a hybrid simulation method for reduced-parameter models updating based on sensitivity analysis is proposed.It incorporates categorical sensitivity analysis and staged model updating to notably reduce the number of parameters analyzed and updated,enhancing parameter identification efficiency.Taking the steel frame model as an example,8 model parameters are selected for analysis.The impact of the hybrid simulation updating in the reduced-parameter model with different sensitivity shares as the core parameter selection criteria is discussed,employing two parameter reduction strategies,respectively.The numerical simulation results indicate that the discrepancy between errors under two reduction strategies is less than 0.5%.Categorical and staged reduction strategies significantly decrease the computational burden of parameter identification and the number of convergence steps in comparison to full parametric strategy.Thus,the feasibility and effectiveness of the proposed method in enhancing the speed of hybrid simulation parameter identification are confirmed.

关 键 词：混合模拟 分类敏感性分析 分阶段模型更新 参数识别 均匀设计 

分 类 号：TU311.4[建筑科学—结构工程]