1、论文中英文摘要作者姓名:曹会兰论文题目:基于地震观测记录的多自由度结构参数识别方法及应用的研究作者简介:曹会兰,女, 1982 年 4 月出生,2005 年 9 月师从于中国地震局工程力学研究所李山有研究员,于 2008 年 8 月获硕士学位。中 文 摘 要结构参数识别是结构抗震安全性能鉴定和健康诊断的基础,利用地震观测记录来识别结构模态参数,是地震工程领域备受关注的研究课题之一。本文基于结构理论地震反应时程和实际观测记录,对多自由度结构模态识别方法及其应用进行了探讨,主要研究内容如下:1. 总结了当前国内外结构参数识别方法的研究成果,介绍了一维、多维和整体外源自回归模型(ARX 模型)以及相
2、应的结构参数识别理论。2. 对激励线性分段插值法和平均加速度法两种地震反应分析方法进行了对比分析,应用激励线性分段插值振型叠加法计算了三层平面框架结构在 El Centro 地震波输入下的理论反应时程。3. 利用三层平面框架结构的理论地震反应时程,对一维 ARX 模型、多维ARX 模型和整体 ARX 模型三种模态参数识别方法进行了对比分析。结果表明:1) 在直接以输入地震波和理论地震反应时程为识别数据时,三种方法具有相同的识别效果且精度很高。2)在以添加噪声后的输入地震波和理论地震反应时程为识别数据时,一维 ARX 模型在不同测点会识别出不同的结果,阻尼比的出入有时会很大;当某测点输出噪声较大
3、或模态耦合严重时,一维和多维 ARX 的识别精度往往会受到较大影响,甚至会得出错误的结果,而整体 ARX 模型的识别结果较为稳定且精度较高。3) 识别参数精度与识别数据中的噪声水平有密切关系,噪声越大识别精度就越差,模型的符合度也越低;噪声对高阶模态的识别影响比低价模态大,对阻尼比的识别影响比频率大。4. 基于 5 个建筑物的多点地震观测记录,应用一维 ARX 模型、多维 ARX模型和整体 ARX 模型,识别了相应的结构模态参数。结果表明:1)整体 ARX模型识别结构参数要比多维 ARX 稳定,多维 ARX 模型在阶数较低时有时候会丢失模态、在阶数较高时又常会产生虚假模态,同时在某一测点的实测
4、值出现异常时多维 ARX 很可能会偏离正确结果;而整体 ARX 模型基本上能有效地识别出相对精确的结果。2)整体 ARX 模型的识别结果多处于一维 ARX 模型的识别结果中间,而多维 ARX 的识别结果常处于一维 ARX 模型的识别结果两侧。5. 对滞回曲线结构参数识别理论进行了讨论,并应用到上述三层平面框架结构的参数识别。通过对比分析发现:1)滞回曲线识别结构参数法能高精度地识别出结构的刚度和阻尼比,且刚度的识别精度比阻尼比高。2)对原始数据进行滤波处理后,刚度和阻尼比的识别精度会有所提高。最后,论文对研究工作进行了总结,对今后研究进行了展望。关键词:结构参数识别;ARX 模型;滞回曲线结构
5、参数识别方法;理论地震反应时程;结构地震动观测记录Parameter Identification Methods and Application of Multi-degree of Freedom Buildings Using Seismic Observation Records Cao HuilanABSTRACTStructural parameter identification is the basis of safety assessment for earthquake-resistance and health diagnosis of structures. Struc
6、tural modal parameter identification using seismic observation records receives wide attention in earthquake engineering. In this paper, the identification methods and application of multi-degree of freedom buildings are discussed using the theoretical time-histories of seismic response and actual o
7、bservation records. Following is the main content:1. Research achievements of current domestic and international structural parameter identification methods are summarized. One-dimensional, multi-dimensional, integral auto-regressive with exogenous input models (ARX model) and corresponding structur
8、al parameter identification theory are introduced.2. Two seismic response analysis methods are compared and analyzed, the piecewise linear interpolation method of excitation and the average acceleration method. The former method is used to calculate the seismic response of a three-story plane frame
9、structure under El Centro wave as bottom input.3. Modal parameter identification methods of one-dimensional, multi-dimensional and integral ARX models are compared and analyzed using the seismic response of the three-story plane frame structure. The results show that: 1) When we directly take the in
10、put seismic wave and the theoretical time-histories of seismic response as identification data, the three methods have the identical identification effect and high precise. 2) When we take the input seismic wave and the theoretical time-histories of seismic response which are added noises as identif
11、ication data, one-dimensional ARX gives different results at different measuring points, and sometimes the damping ratio varies greatly. When some measuring points mix with big noises or the mode coupling is serious, the influence on the precise of one-dimensional ARX and multi-dimensional ARX model
12、s are great, even giving false results, while integral ARX model gives stable and highly precise results. 3) The precision of parameter identification is closely related to the level of noises in identification data. The bigger the noises, the poorer the precision of results and the conformity accur
13、acy. The influence on high order modes is larger than the low order modes, and the influence on damping ratio is larger than the frequency.4. These three ARX methods are applied to the multi-point seismic observation records of five buildings to identify the corresponding structural modal parameters
14、. The results show that: 1) The integral ARX model is more stable than the multi-dimensional ARX model for the latter taking the coupling influence into consideration, sometimes the results of which will be mode loss, and sometimes will give false mode by using bigger order. At the same time, when t
15、he record of certain measuring point is abnormal, multi-dimensional ARX model may give false results while the integral ARX model gives the relatively effective precise results. 2) Most estimation results of integral ARX model are between the results given by one-dimensional ARX model, while those o
16、f multi-dimensional ARX model are always beyond the results given by one-dimensional ARX model.5. Parameter identification theory using the hysteresis curve is discussed, which is applied to the parameter identification of above three-story plane frame structure. According to the comparison and anal
17、ysis, it shows that: 1) The structural parameter identification method using the hysteresis curve can estimate the stiffness and damping ratio with high precise, as the precise of stiffness is higher than that of damping ratio. 2) The precise of the stiffness and damping ratio estimation results is
18、basically improved after the filtering of the original data.At last, the research work is summarized and the prospects of the future research work are described in this paperKey words: structural parameter identification; ARX model; structural parameter identification method using hysteresis curve; theoretical time-history of seismic response; seismic observation records of building;