| 引用本文: | 张石磊,陈少峰,王焕定,王伟.基于正演分析的模型参数修正[J].哈尔滨工业大学学报,2012,44(10):1.DOI:10.11918/j.issn.0367-6234.2012.10.001 |
| ZHANG Shi-lei,CHEN Shao-feng,WANG Huan-ding,WANG Wei.Model parametric updating based on forward analysis[J].Journal of Harbin Institute of Technology,2012,44(10):1.DOI:10.11918/j.issn.0367-6234.2012.10.001 |
|
| 摘要: |
| 为探索测量信息不完整情况下基于均匀设计的正演分析法的适用性,设计并修正了一个钢桁架模型．在获得少量测量数据的情况下,分析和确定了主要修正因素．利用变截面梁模拟了钢桁架节点的力学特性．采用置信区间确定了识别结果的取值范围．采用全因素、嵌套和分步3种均匀设计方案修正了钢桁架模型参数．计算结果显示:采用上述方法确定的模型修正因素及其水平正确有效,基于分步均匀设计方案的正演分析法的修正结果最合理．在实际结构的模型修正中,应正确确定模型修正因素并分清主次,构建小规模均匀设计方案、分多步修正模型,以减少计算量和避免不合理的模型修正解．基于研究过程中的教训与经验,最终总结出了一套适用于大型复杂结构模型修正的方法并给出计算步骤． |
| 关键词: 模型修正 正演分析 均匀设计 神经网络 主因素 |
| DOI：10.11918/j.issn.0367-6234.2012.10.001 |
| 分类号:TU317 |
| 基金项目:国家自然科学基金资助项目(50978078). |
|
| Model parametric updating based on forward analysis |
|
ZHANG Shi-lei1, CHEN Shao-feng2, WANG Huan-ding1, WANG Wei1
|
|
1.School of Civil Engineering, Harbin Institute of Technology, 150090 Harbin, China;2.School of Transportation Science and Engineering, Harbin Institute of Technology, 150090 Harbin, China
|
| Abstract: |
| To study the applicability of the forward analysis method based on uniform design under incomplete metrical information, a steel truss is designed and its model parameters are updated. The principal factors of model updating are analyzed and determined on condition that a few of data is measured. The mechanical characteristics of the steel truss joints are simulated by a kind of beam with variable cross-section. The value range of identification results is determined by confidence interval. Three kinds of uniform design program, which are the full-factor program, the nesting program and the multistep program, are used to correct the design parameters of the steel truss. The results show that the model updating factors and their levels determined by above methods are accurate and effective, and the results of model updating by the multistep uniform design program is the most reasonable one. When the mechanical model of a real structure is updated, the factors of model updating should be determined accurately and draw a clear distinction between the primary and the secondary. Multi small-scale uniform design programs should be constructed. And the model should be updated with multistep using these small-scale programs to reduce workloads and avoid irrational model updating solutions. At last, based on the lessons and experiences of the study, a set of methods and its calculation procedure, which are suit for model updating of large complicated structures, are summarized and described. |
| Key words: model updating forward analysis uniform design neural network principal factor |
