环境振动模态参数识别及其在土木工程结构中的应用

环境振动模态参数识别及其在土木工程结构中的应用

任伟新博士

福州大学土木建筑工程学院

特聘教授，博士生导师

中南大学桥梁与隧道工程博士生导师

Bridge Stability and Dynamics Lab 91fcc226482fb4daa58d4bae Faculty Members:

?Dr. Wei-Xin Ren （任伟新）?Dr. Zhouhong Zhou （宗周红）

Bridge Stability and Dynamics Lab Student Members:

? 1 post-doctoral researcher ? 6 full time Ph.D. students ?15 full time M.Sc. students

91fcc226482fb4daa58d4bae

What Are We Doing??Nonlinear finite element method and applications;?Long span cable-supported bridges;

?Seismic analysis and design of structures;?Dynamic testing and system identification;?Damage detection and diagnosis of structures;?Structural evaluation, strengthening and long-

term heath monitoring.

?Composite Bridge and Structures.

91fcc226482fb4daa58d4bae

Ph.D. Topics

?Finite element model updating through ambient vibration measurements (Mr. Bijaya Jaishi);

?Damage identification through the changes in structural dynamic characteristics (Mr. Sun Zengshou);

?Ambient vibration based system identification of bridges and applications (Mr. Yu Danjiang);

?Statistics based method of bridge on-line health monitoring (Mr. Lin Youqing);

?Ultimate load-carrying capacity analysis of complicated bridges;?Nonlinear analysis of concrete boxed-section bridges

?Nonlinear cyclic behavior of connections.

91fcc226482fb4daa58d4bae

系统识别

系统识别的含义：工程结构系统[M]，[C]，[K]输入

输出激励响应传统系统识别的不足：?需要专用的系统激励设备；?必须封闭线路，无法实现实时的安全监测；

土木工程结构动力试验方法

强迫振动试验法;

自由振动试验法;

环境振动试验法。

激励方式

力锤模拟正常交通车辆

斜拉桥自由振动测试

Problems with Civil Engineering Structures ?It is extremely difficult to realize the excitation on a large-scale structure. Some heavy forced excitations become very expensive.

?Traffic has to be shut down for a rather long time. This could be a serious problem for intensively used bridges.

?The need to identify modal models under operational conditions often arises for the on-line monitoring.

91fcc226482fb4daa58d4bae

环境（自然）激励的优点

天然、方便和便宜和的激励方式； 更符合实际情况和边界条件；

可以实现对结构的实时安全监测；

环境激励的缺点

结构动力响应测试数据，具有幅值小、随机性强的特点；

记录时间长，数据量巨大；

系统识别是仅由输出数据的系统识别方法(Output-only System Identification) ；

给结构系统的识别带来很大的难度

Output-Only SI Methods

?Peak-picking from power spectral densities (PSDs);

?Auto Regressive-Moving Average (ARMA) model based on discrete-time data;

?Natural excitation technique (NExT);

?Stochastic subspace methods et al.

There have been several ambient vibration SI techniques available that were developed by different investigators or for different uses such as:

91fcc226482fb4daa58d4bae

存在的问题

远还不能够说是一个已经完全解决了的课题； 数据减缩、方程求解法和矩阵运算的顺序等；

大型复杂工程结构系统识别算法和计算机GUI 实现；

在全尺寸实际结构的应用方面;

Frequency-Domain SI Techniques ?Based on the fact that FRFs goes through an extreme around the natural frequencies.

?The reference signal is used as an "input" and FRFs and coherence functions are computed for each measurement point with respect to reference point.

?The most popular, mainly due to their simplicity and processing speed, and also for historical reasons. ?Involve averaging temporal information, thus discarding most of their details;

?Always a real modal analysis.

91fcc226482fb4daa58d4bae

Time-Domain SI Techniques ?Directly work with time data, without the need to convert them to correlations or spectra;

?Based on a discrete-time state space model of a dynamic system;

?Identify the state space matrices based on the measurements and then determine modal parameters;?Need robust numerical techniques such as QR-factorization or least squares;

?It is always complex modes.

91fcc226482fb4daa58d4bae

Peak-Picking in Frequency-Domain ?The simplest approach to estimate the modal parameters of a structure subjected to ambient loading.

?The FRFs are simply replaced by the auto spectra of the ambient outputs without the real FRFs computed.

?the natural frequencies are simply determined from the observation of the peaks of the average normalized power spectral densities (ANPSDs) obtained by a discrete Fourier transform (DFT).

?Mode shapes are simply replaced by operational deflection shapes.

91fcc226482fb4daa58d4bae

Stochastic Subspace Identification in Time-Domain ?Input is replaced by process noise w k and measurement noise v k in discrete-time state space model of a dynamic system;

?These noises are assumed to be the white noise;

?The QR factorization results in significant data reduction;?The singular value decomposition (SVD) is used to reject the noise (represented by the smaller singular values);?Once mathematical state space model is identified, it is straightforward to determine the modal parameters (by an eigenvalue decomposition).

91fcc226482fb4daa58d4bae

Output-Only Modal Analysis: A GUI for MATLAB ? A Graphical User Interface (GUI) for output-only modal analysis is developed;

?The simple peak-picking method as well as the more advanced stochastic subspace method are implemented in

a user-friendly way.

?By pushing buttons the user is guided through the whole process of output-only modal analysis: converting measurements to engineering units, preprocessing the data, system identification, gluing mode shape parts together, animating mode shapes, . . .

91fcc226482fb4daa58d4bae

Main Window-Three Main Tasks

91fcc226482fb4daa58d4bae

本文来源：https://www.bwwdw.com/article/869q.html