基于截面屈服面模型更新的框架结构混合试验方法

发布时间：2018-01-23 14:37

本文关键词： 混合试验本构模型系统辨识参数估计模型更新　出处：《哈尔滨工业大学》2015年博士论文　论文类型：学位论文

【摘要】：混合试验综合了数值模拟和物理试验的优点,是研究结构动力反应的有效方法之一。子结构技术的应用使混合试验在评估大型复杂结构抗震性能上具有很大的潜力。但是当前的子结构混合试验还存在较大模型误差:一方面源自结构的计算模型和数值单元过于简化,如用层模型及构件层次本构来描述框架结构;另一方面源于数值子结构模型参数的不确定性,如用假定的数值模型来描述大型复杂结构中无法进行试验又可能进入非线性的构件,当这种数值模型在结构中所占比例较大时,混合试验的优势就消失了。为了减少子结构混合试验中的两类模型误差,有必要采用合理的计算模型和精细的数值单元;可以而且应该利用试验子结构观测数据实时辨识预设数值模型的参数,并在线更新数值子结构中相应部分的模型。这两个解决方案还要求有基于有限元软件的混合试验系统提供技术支撑。本文以框架结构为研究对象,以减少模型误差提高试验精度为宗旨,提出了基于截面屈服面模型更新的框架结构混合试验方法,主要内容和结论如下:(1)提出了梁柱单元中考虑轴力弯矩耦合的截面本构模型。根据截面组合思想构建了轴力与弯矩的屈服面,并基于经典塑性理论建立了截面本构模型;阐述了截面状态确定和本构关系积分的方法;编写了基于MATLAB的杆系有限元非线性分析程序。研究表明,该截面本构模型不但能考虑轴力存在及变化对截面抗弯承载力的影响,而且具有随动、等向及混合强化的特性。此外,与塑性铰模型和纤维截面模型结果的比较表明所提截面模型能兼顾计算精度和效率。(2)提出了梁柱单元中截面本构模型参数的识别方法。以框架柱为辨识对象,提取其基本坐标系下杆端位移和力为输入和输出信息,并预设基于力的单元和截面本构为框架柱的数值模型;采用unscented Kalman Filter(UKF)纯参估计和双重估计模式进行辨识;分析了UKF纯参数估计性能对初值与噪声的选择的敏感性,并探讨了有无模型误差下两种估计模式的识别效果。研究表明,UKF算法在识别梁柱截面本构模型参数时具有较高的精度、鲁棒性和计算效率。(3)建立了基于有限元并考虑模型更新的混合试验系统架构。分析了两环混合试验系统架构存在的问题并搭建了一个混合试验系统仿真平台;研究了基于有限元软件的三环混合试验系统架构,总结了MATLAB有限元与中间环d SPACE间数据的通信,扩展了中间层d SPACE以包含命令的插值、边界的控制等功能;提出面向单元并考虑在线模型更新的杆系结构混合试验系统架构。研究表明,所提出的仿真平台和试验架构是可行的,有效的。(4)开展了常规混合试验及考虑在线模型更新的混合试验。以平面钢框架为研究对象,以杆系结构为计算模型并明确了静动力分析算法;设计了足尺框架柱试验子结构,研究了三个作动器加载模式和测量方案;开展了面向单元的框架结构常规混合试验和考虑在线模型更新的混合试验。研究表明,三个作动器加载模式能实现平面框架柱的边界条件,常规子结构混合试验对大型框架结构优势不太明显,而考虑在线模型更新的混合试验能大幅提高试验精度。
[Abstract]:Hybrid testing combines the advantages of numerical simulation and physical experiment, is one of the effective methods to study dynamic response of structure. The application of Substructure Hybrid Test in the assessment of potential seismic performance of large complex structures. But the Substructure Hybrid test current is still large model error: on the one hand from the model and numerical simulation the unit structure is too simplified, such as the use of layer model and component level constitutive description of frame structure; on the other hand from the numerical substructure model parameter uncertainties, such as the numerical model is assumed to describe the component test can not be carried out in large and complex structures may enter the nonlinear, when a larger proportion of the numerical model accounted for in the structure, the hybrid testing advantage disappears. In order to reduce the model error of two kinds of Substructure Hybrid test, it is necessary to adopt reasonable calculation model And the numerical unit can and should be fine; by using the parameters of the experimental substructure observation data real-time identification of preset numerical model, and the online update corresponding part of the numerical substructure model in two. This solution also requires a hybrid test system based on finite element software to provide technical support. Based on the frame structure as the research object, in order to reduce the error of the model for the purpose of improving the accuracy of the test, put forward a framework of hybrid testing method based on the updated section yield surface model, the main contents and conclusions are as follows: (1) proposed constitutive model of section beam column element considering axial force bending moment coupling of the senior high school entrance examination. According to the section combined to construct the yield surface of axial force and bending moment and, based on the constitutive model of section established the classical plasticity theory; expounds the section state determination and constitutive relation of integral method; preparation of finite element non line based on MATLAB Analysis of the program. The results show that the section constitutive model can not only consider the existence and change of the axial force effect of flexural bearing capacity of cross section, and with follow-up, to strengthen and mixed characteristics. In addition, with the plastic hinge model and fiber section model results show that the proposed model can take into account the section the accuracy and efficiency of computation. (2) proposed recognition method of constitutive model parameters of beam and column in the unit. The frame column for the identification of object extraction, the basic coordinate rod end displacement and force for the input and output information, and the default unit and section based on stress constitutive numerical model for frame columns; using unscented Kalman Filter (UKF) pure parameter estimation and double estimation model identification; analysis of UKF pure parameter estimation and noise performance sensitivity to initial value selection, and discusses a model error of two kinds of estimation model. Research on recognition effect Show that the model has a high precision parameter constitutive UKF algorithm in recognition of beam and column section, robustness and computational efficiency. (3) established the finite element model and considering the hybrid test system architecture based on the updated. Analysis of the existing two ring hybrid test system architecture and build a hybrid test system simulation platform study on the architecture of three; hybrid test system based on finite element software, summarizes the MATLAB finite element and the intermediate ring D data communications between SPACE and extended middle layer D SPACE to include command interpolation, boundary control; the unit and consider the online model updating bar structure hybrid test system architecture. The research shows that the simulation platform and the test architecture proposed is feasible and effective. (4) to carry out routine mixture test and hybrid test consider online model updating. The plane steel frame for the research of As to truss structure calculation model and the static and dynamic analysis algorithm; design of full-scale column test structure, studied three actuator loading mode and measurement scheme; carry out frame structure of conventional hybrid test oriented unit and considering the mixed test online update model. The results show that, three the actuator loading mode can realize planar frame column boundary conditions, conventional Substructure Hybrid Test of large frame structure advantage is not obvious, and consider the online hybrid test model updating can greatly improve the test precision.

【学位授予单位】：哈尔滨工业大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：TU317.1;TU311.3

【参考文献】