基于模型修正与灰色系统理论的梁桥损伤识别研究
Research on damage identification of girder bridge based on model correction and gray system theory
周宇 1石英迪 2李萌 2孙文卓 2张德伟2
作者信息
- 1. 安徽建筑大学 土木工程学院,安徽 合肥 230601;兰州交通大学 土木工程学院,甘肃 兰州 730070;安徽建筑大学 建筑健康监测与灾害预防技术国家地方联合工程实验室,安徽 合肥 230601
- 2. 安徽建筑大学 土木工程学院,安徽 合肥 230601
- 折叠
摘要
对既有梁桥结构开展损伤识别分析时,需要依托有限元软件建立理想的梁桥数值模型,然而依据设计图纸建立的梁桥有限元模型尚未考虑到建模过程的不确定性,有限元模型计算与既有梁桥实测存在较大偏差.利用基于遗传算法优化BP神经网络的模型修正方法,对修正后更加贴近实际梁桥结构的有限元模型展开数值分析,通过挠度影响线进行灰色系统相关性分析与损伤识别.修正结果表明:基于遗传算法优化BP神经网络的梁桥模型修正方法可使有限元模型贴近实际结构模型;结合现场模态测试,结果频率误差从未修正的 30.5%降至修正后的9.8%.损伤识别结果表明灰色系统理论相关系数与损伤程度呈线性关系,灰色系统相关系数越小,损伤程度越大.
Abstract
In order to carry out the damage identification analysis of existing girder bridge structure,it is necessary to rely on finite element software to establish an ideal bridge numerical model.However,the finite element model of the girder bridge established based on the design drawings has not yet taken into account the uncertainty of the modeling process,and the calculation of the finite element model has a large deviation from the actual measurement of the existing bridge.Using the model modification method based on BP neural net-work optimized by genetic algorithm,numerical analysis is carried out on the modified finite element model which is closer to the actual girder bridge structure,and gray system correlation analysis and damage identifi-cation are carried out through the deflection influence line.The results show that the proposed model modifica-tion method based on genetic algorithm optimized BP neural network makes the finite element model close to the actual structural model,and combined with the field modal test,the resultant frequency error decreases from the uncorrected 30.5%to the corrected 9.8%.The damage identification results show that the gray sys-tem theory correlation coefficient is linearly related to the damage degree,and the smaller the gray system cor-relation coefficient is,the greater the damage degree is.
关键词
梁桥结构/模型修正/损伤识别/遗传算法/灰色系统理论Key words
bridge engineering/model correction/damage identification/genetic algorithm/gray sys-tem theory引用本文复制引用
基金项目
安徽省高校科学研究重点项目(2022AH050248)
甘肃省建设科技项目(JK2023-03)
企业委托技术开发课题(HYB20220240)
企业委托技术开发课题(HYB20230001)
建筑健康监测与灾害预防技术国家地方联合工程实验室主任基金(GG22KF002)
出版年
2024
NETL
NSTL
维普
万方数据