Two-stage offline-online FE model updating with uncertainty
quantification for cable-stayed bridges
Hong Li,Kun Zhang,Zhouhong Zong,Jie Niu
Abstract
Online model updating of bridge structures during operation represents a critical challenge in structural health monitoring. A two-stage online model updating strategy for long-span cable-stayed bridges is developed using the Guanhe Bridge as a case study. In Stage 1, a reference finite element model accounting for parameter uncertainties is established through a sensitivity-based interval response surface method with historical offline data. Sensitivity analysis, optimal Latin hypercube sampling, and multi-objective optimization via a multi-island genetic algorithm are integrated. An average relative error within 0.5% is achieved for the target frequencies. The resulting model captures the uncertainty range of operational parameters and serves as a reliable benchmark for online updating. In Stage 2, online-identified frequency information is incorporated into a deterministic model updating process, enabling rapid dynamic model updates. This finite element model updating, based on online modal frequencies, yields a maximum relative error of 0.86% for the second horizontal mode H2. The uncertainty of the updated parameters is substantially reduced, and the credibility of the results is enhanced. Dynamic model updates are thus achieved on a shorter time scale. This two-stage method provides theoretical foundations and technical support for online safety state assessment of cable-stayed bridges.
Hong Li, Kun Zhang, Zhouhong Zong — Engineering Research Center for Safety and Protection against Explosion and Impact of Ministry of Education, School of Civil Engineering, Southeast University, Nanjing, Jiangsu, 211189, China
Jie Niu — School of Civil Engineering and Architecture, Nanjing Institute of Technology, Nanjing, Jiangsu, 211189, China
PDF Viewer
Preview is limited to the first 3 pages. Sign in to access the full PDF.
