| 摘要: |
| 渡槽在服役过程中，由于结构会受到各种因素的影响，其结构状态会发生改变，所以在进行有限元分析时，建模所用参数并不是结构实时参数。因此，需要通过监测结构的动力响应来对结构的有限元模型参数进行更新，达到准确评价结构实际运行状态的目的。本文提出了一种基于贝叶斯理论的有限元模型更新方法，该方法以确定性模型更新结果作为不确定性更新的初始模型参数值，使用MH算法得到模型参数的后验分布，通过ABAQUS有限元模型与MATLAB更新程序的交互访问，实现复杂结构有限元模型参数的自动更新。开展了渡槽模型试验，利用所提方法对渡槽的有限元模型进行更新。结果表明，以确定性更新的结果作为贝叶斯更新的参数初始值，渡槽模型有限元模型更新的计算效率提高了25%，更新之后的误差在0.2%到7%之间，说明该更新方法能够有效地改善有限元模型的预测能力。 |
| 关键词: 模型更新 贝叶斯理论 MH算法 随机子空间法 |
| DOI： |
| 分类号: |
| 基金项目:水利青年人才发展资助项目（SS01990662022），中国水利水电科学研究院五大人才计划项目（SD0145B042021），流域水循环模拟与调控国家重点实验室自主研究课题（SKL2020ZY10） |
|
| A Finite Element Model Updating Method for Aque-duct Structures Based on Bayesian Theory |
|
Wu Xijie1, Zhang Yan2, Yin Jinbu1, He Junling1, Zhang Yao1
|
|
1.College of Water Resources and Architectural Engineering, Northwest Agricultural and Forestry University;2.State Key Laboratory of Simulation and Regulation of Water Cycle in River Basin, China Institute of Water Resources and Hydropower Research
|
| Abstract: |
| During the service process of the aqueduct, the structure will be affected by various factors, and its structural state will change. Therefore, in the finite element analysis, the parameters used for modeling are not real-time parameters of the structure. Therefore, it is necessary to update the finite element model parameters of the struc-ture by monitoring the dynamic response of the structure to achieve the purpose of accurately evaluating the actual operating state of the structure. In this paper, a finite element model updating method based on Bayesian theory is proposed. The method uses the deterministic model update result as the initial model parameter value of the uncertainty update, and uses the MH algorithm to obtain the posterior distribution of the model parameters. Through the interactive access between the ABAQUS finite element model and the MATLAB update program, the automatic update of the finite element model parameters of the complex structure is realized. The aqueduct model test was carried out, and the finite element model of the aqueduct was updated by us-ing the proposed method. The results show that the calculation efficiency of the finite element model updating of the aqueduct model is improved by 25 %, and the error after updating is between 0.2 % and 7 %, which indicates that the updating method can effectively improve the prediction ability of the finite element model. |
| Key words: Model updates Bayesian theory MH algorithm Random subspace method |
