%0 Journal Article
%T 基于UKF的结构动荷载识别方法与试验验证
%A 郭丽娜
%A 宋开明
%A 张延哲
%A 丁勇
%A 
%J 振动与冲击
%D 2019
%X 结构荷载识别是结构动力系统反问题研究的一个重要方面，而实际工程结构动力荷载的有效识别对结构设计、维护和加固具有重要的指导意义。本文首先利用基于UKF算法的两种荷载识别方法，即基于荷载时程正交分解的间接荷载识别法和直接荷载识别法，对一单自由度滞回非线性剪切模型进行数值仿真模拟，验证了直接荷载识别法应用于单自由度非线性模型荷载识别的合理性和有效性。随后，对抗震实验室振动台进行空台加载，通过无线传感器测量振动台台面动力响应，利用UKF算法识别地震模拟振动台的刚度和阻尼参数。在已识别参数的基础上，采用直接荷载识别法对外部激励进行识别，并针对出现漂移现象的识别结果提出了位移-加速度联合观测值和噪声水平自适应评估两种直接扩展状态量的荷载识别方法，改善了识别过程的收敛性。</br>Abstract：Load identification is an important aspect in inverse problem research of structural dynamic systems.The effective identification of structural dynamic load can provide an important guide for structure design, maintenance and reinforcement in practical engineering.Here, two load identification methods based on UKF algorithm including the indirect load identification method based on the orthogonal decomposition of load time history and the direct load identification one were used to do numerical simulation for a single-DOF hysteretic nonlinear shear model to verify the rationality and effectiveness of the direct load identification method applied in the load identification of the single-DOF nonlinear model.Firstly, an empty shaking table was loaded and its table surface’s dynamic response was measured with a wireless sensor.Then, stiffness and damping parameters of the table were identified with UKF algorithm.After recognizing all parameters, the external load was identified using the direct load identification method.Aiming at the recognized results with “drift” phenomena, two methods to directly extend state variables including the joint observation of displacement and acceleration, and the noise level adaptive evaluation were proposed to improve the convergence of the recognition process.
%K 动荷载识别
%K UKF算法
%K 地震模拟振动台
%K 无线传感器
%K 漂移&lt
%K /br&gt
%U http://jvs.sjtu.edu.cn/CN/abstract/abstract8233.shtml