Parameter Variability Estimation by Stochastic Model Updating
Traditional FE model updating methods are generally deterministic since uncertainties existing in param eters and measurements are not considered.However, for most real-world structures, uncertainties caused by inherent parameter variability, external environmental factors and measurement noises always exist and thus cannot stand apart from analysis.Therefore, a stochastic model updating procedure involving the uncertainty analysis should be performed in order to identify, instead of deterministic values, the statistical properties of parameters.This study attempts to quantify parameter variability resulting from geometric tolerances and ma terial discreteness.For simplification, the stochastic model updating process is decomposed into a series of deterministic ones where response surface models have been used as surrogates for original FE models for the purpose of cost-efficiency in programming and computation.Monte Carlo simulation is employed to generate response samples each of which corresponds to one deterministic updating process and thus a set of parameter values can be predicted.Lastly parameter means and standard deviations are estimated based on all the sam ple predictions.The proposed method has been validated by a set of nominally identical metal plates experi mentally tested.And the updating results demonstrate the feasibility and efficiency of the proposed method.
Shengen Fang Qiuhu Zhang Youqin Lin
School of Civil Engineering, Fuzhou University, Fuzhou, China School of Civil Engineering, Hefei University of Technology, Hefei, China
国际会议
哈尔滨
英文
695-701
2013-07-06（万方平台首次上网日期，不代表论文的发表时间）