基于模型更新的土木结构混合试验方法

发布时间：2018-03-19 14:37

本文选题：混合试验　切入点：模型更新　出处：《哈尔滨工业大学》2014年博士论文　论文类型：学位论文

【摘要】：结构混合试验是将物理加载试验和数值模拟相结合来评估大型复杂土木结构地震反应和抗震性能的有效试验技术，目前正得到了研究者们的广泛关注。目前已在混合试验系统、数值积分算法、加载控制、时滞补偿、误差累计控制等关键技术方面取得了一定的研究成果。然而，对于高层建筑或者大跨度桥梁结构进行混合试验时，研究者们会遇到很大困难。当整体结构进入非线性，不可能将所有关键部分都进行物理试验，如何准确模拟数值子结构成为最大的挑战。为了能够提高混合试验中数值子结构模型的准确性，提出了基于约束隐性卡尔曼滤波器的模型更新混合方法。模型更新混合试验方法的研究对推动混合试验方法发展及客观揭示大型复杂土木结构的动力灾变过程具有重要的理论意义和实用价值。本文针对基于模型更新的土木结构混合试验方法的一些关键科学问题进行了研究，主要研究内容如下： (1)采用确定性最小二乘法在线识别双折线模型参数，通过数值模拟和试验验证了在相同加载路径下的在线模型更新效果。研究表明：该方法具有良好的参数识别精度和较高的计算效率。针对两层防屈曲支撑结构，首次进行了模型更新的混合试验验证。试验研究表明：与传统混合试验相比，在线模型更新可以有效地减小混合试验数值模型误差，提高混合试验精度。 (2)针对常量标量系统，推导了不考虑过程噪声和考虑过程噪声状态估计方差解析表达式，给出了卡尔曼滤波器影响参数对状态估计影响规律。采用状态估计方差作为状态估计评价标准，给出了使用卡尔曼滤波器进行状态估计时滤波器参数取值建议。验证了采用状态投影方法来解决界限约束状态估计的合理性。提出了考虑乘性噪声卡尔曼滤波器的最优估计递推算法，通过弹性结构刚度识别验证了算法的有效性。 (3)研究了基于隐性卡尔曼滤波器(UKF)的模型更新方法，，搭建了基于dSPACE控制平台的FTS作动器试验系统，针对弹性试件首次进行了基于UKF的模型更新混合试验。试验表明：与传统混合试验相比，该试验方法大幅度地提高了混合试验精度。提出了基于修改Sigma点(Sigma Points)位置及权重的约束隐性卡尔曼滤波器算法(约束UKF)。与UKF相比，约束UKF算法主要有两个方面的改进：1)在预测步中，将违反界限约束的Sigma点移至边界上，同时将约束界限内的Sigma点做相应对称调整，并从理论上证明当κ=0.5时考虑约束的UT变换算法具有一阶精度；2)在校正步中，采用状态更新方程产生变换后的Sigma点，当更新的状态估计违反界限约束时，直接将超出约束边界的Sigma点投影到边界上，并给出了校正步状态估计均值和协方差计算公式的理论证明。所提出的算法可以同时考虑了界限约束对状态估计均值和协方差的影响，而且具有较高的计算效率，能够实现实时模型更新。给出了Bouc-Wen模型在线参数识别方法。研究表明：与UKF方法相比，约束UKF方法能够有效地减小参数识别值前期波动幅度，确保参数识别值的物理意义，提高参数识别收敛速度和模型更新精度。进行了滤波器初始参数影响分析，并给出参数选择建议。 (4)针对两个自由度非线性结构进行了基于约束UKF模型更新的混合试验数值模拟，验证了该试验方法的有效性和可行性。采用dSPACE-MTS实时混合试验系统，分别针对防屈曲支撑结构首次进行了慢速和实时的模型更新混合试验。试验表明：与传统混合试验和基于UKF模型更新的混合试验相比，基于约束UKF模型更新的混合试验方法可以更大程度地提高混合试验精度。 (5)提出基于OpenSees的模型更新混合试验方法，针对一榀六层框架-支撑结构，进行了多种情况下的模型更新混合试验数值仿真。数值子结构和试验子结构均采用OpenSees进行数值模拟，模型参数识别采用Matlab编程实现。通过修改原OpenSees程序源代码，建立可以模型更新的twoNodeLink单元。采用Socket通信技术实现OpenSees和Matlab程序的数据传输。分别讨论试验支撑采用Bouc-Wen模型和考虑强度、刚度退化的Bouc-Wen模型两种情况下模型误差对混合试验精度的影响。研究表明，模型误差将导致模型参数识别值很难收敛到一个稳态值，具有较强的时变性，这会降低模型更新混合试验精度。提出了将模型更新与多个试验子结构混合试验相结合来减小模型误差影响的方法，尝试选择多个关键部位的构件作为试验子结构，然后分别更新与其加载路径和非线性程度相近的数值子结构模型。研究表明，该方法能够有效地减小模型误差对模型更新的影响，提高了模型更新混合试验精度。
[Abstract]:The structure of hybrid test is the physical load test and numerical simulation were combined to evaluate the effective test technology of structural seismic response and seismic performance of complex civil engineering, currently has received wide attention from researchers. Now in the hybrid test system, numerical integral algorithm, load control, delay compensation, the key technology of cumulative error control has made some achievements. However, for high-rise buildings or large span bridge structure of hybrid test, researchers will encounter great difficulties. When the overall structure in nonlinear, can not be all of the key part of the physical test, how to accurately simulate the numerical substructure has become the biggest challenge. In order to improve the accuracy of the numerical the sub structure model of mixed test, the proposed constraint implicit Calman filter model updating method based on mixed model. The mixed update test The research of inspection method is of great theoretical and practical value for promoting the development of hybrid test method and objectively revealing the dynamic catastrophe process of large complex civil structures.
In this paper, some key scientific problems in the hybrid test method of civil structure based on model updating are studied. The main contents are as follows:
(1) using deterministic least squares on-line identification of bilinear model parameters through numerical simulation and experimental verification of the online model in the same loading path update effect. The calculation results show that the efficiency of this method has good accuracy and high parameter identification. According to the two layer of BRB structure, the first hybrid test model update. Test results show that compared with the traditional hybrid test, online update model can effectively reduce the mixed numerical model test error, improve the accuracy of hybrid test.
(2) according to the constant scalar system is derived without considering the process noise and the noise variance of state estimation considering the process of analytic expression gives Calman filter parameter effects on state estimation are studied. The state estimation variance as the evaluation criteria adopted for state estimation, given the use of Calman filters for state estimation filter parameters are verified by the state. The projection method to solve the reasonable limit constrained state estimation is proposed. The optimal multiplicative noise Calman filter estimation recursive algorithm, the elastic stiffness of the structure to identify the validity of the algorithm.
(3) the hidden Calman filter (UKF) based on the model updating method, set up the dSPACE control platform based on FTS actuator test system for elastic specimen for the first time UKF model updating based on the test of mixed. The experimental results show that compared with the traditional hybrid test, this test method greatly improves mixing the accuracy of the test. Put forward the modification based on Sigma (Sigma Points) algorithm and Calman filter constraints implicit position weight (constraint UKF). Compared with UKF, two improvements are the main constraint UKF algorithm: 1) in the prediction step, will break the limits of Sigma point to the boundary, at the same time constraints the Sigma do the corresponding symmetric adjustment, and it is proved that when considering UT K =0.5 transform algorithm with the constraints of the first order accuracy; 2) in the correction, the state update equation produced change after Sigma, when more A new state estimation in violation of limits, directly will exceed the projection constraint boundary point Sigma to the boundary, and gives the correction step state estimation formula proved mean and covariance theory. The proposed algorithm can simultaneously consider the boundary constraints on the effect of state estimation of mean and covariance, and with high computational efficiency. To achieve the real-time model updating. The Bouc-Wen model of online parameter identification method. The results show that, compared with UKF method, constrained UKF method could effectively reduce the value of parameter identification pre fluctuations, to ensure the physical meaning of the parameters identification value, improve the convergence speed of parameter identification and model updating accuracy are analyzed. The initial parameters of the filter effect. And gives the parameter selection suggestions.
(4) for the two degree of freedom nonlinear structure by numerical simulation of mixing test update constraints based on UKF model, verify the validity of the test and the feasibility of the method. The test system for real-time hybrid dSPACE-MTS, respectively for the anti buckling support structure for the first time in the slow and real-time model updating hybrid test. The results show that, compared with the the traditional hybrid test and mixed test UKF model updating based on hybrid testing method update constraints based on UKF model can greatly improve the accuracy of the mixed test.
(5) OpenSees model is proposed to update the test method based on mixed, according to a six storey braced frame structure, a variety of situation model updating hybrid test numerical simulation. The numerical substructure and the experimental substructure were numerically simulated by OpenSees model, parameter identification using Matlab programming by modifying the original OpenSees program. The source code, can establish twoNodeLink cell model updating. The realization of OpenSees data transmission and Matlab program using Socket communication technology are discussed. The test is supported by Bouc-Wen model and considering the strength, stiffness of the back effects of Bouc-Wen model of the two model error on the accuracy of the mixed test. The results show that the model error will lead to the model parameters the identification value is difficult to converge to a steady value, has strong time-varying, which will reduce the accuracy of model updating is proposed. The hybrid test model updating With multiple test Substructure Hybrid test combination method to reduce the influence of model error, try to choose a number of key component parts as experimental substructure, and then update the numerical substructure model with nonlinear loading path and similar levels. The results show that the method can effectively reduce the influence of model error on model updating to improve the model accuracy, update the hybrid test.

【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】：2014
【分类号】：TU317

【参考文献】