结构物理参数时域识别的振动台试验研究*

(1.中国地震局工程力学研究所 中国地震局地震工程与工程振动重点试验室,黑龙江 哈尔滨 150080; 2.广州大学 土木工程学院,广东 广州 510006; 3.广州大学 工程抗震研究中心,广东 广州 510405)

振动台试验; 结构物理参数识别; 地震动反演; 复合反演算法; 非线性参数系统

Research on Shaking Table Test on Structural Physical Parameter Identification in Time Domain
WANG Xiangjian1,CUI Jie2,3

(1. Key Laboratory of Earthquake Engineering and Engineering Vibration,Institute ofEngineering Mechanics,CEA,Harbin 150080,Heilongjiang,China)(2. Department of Civil Engineering,Guangzhou University,Guangzhou 510006,Guangdong,China)(3. Earthquake Engineering Research and Test Centre,Guangzhou University,Guangzhou 510405,Guangdong,China)

shaking table test; structural physical parameter identification; ground motion inversion; hybrid inversion method; nonlinear parametric system

备注

在输入未知条件下,为了识别非线性参数系统,并消除噪声异常点的影响,引入修正Levenberg-Marquardt法和矩形窗法,建立SVD-mLM法,改进了复合反演算法。为了验证改进算法,设计一个五层单跨钢框架模型进行振动台试验,测试获得含有真实噪声的动力响应,采用改进的复合反演算法,研究了结构物理参数时域识别和输入反演问题。振动台试验研究表明结构物理参数识别和输入反演结果是可信的,验证了SVD-mLM法、矩形窗法和改进的复合反演算法在识别结构物理参数和反演输入时的可行性和有效性。

To identify the nonlinear parametric system without the inputting information and eliminate the influence of abnormal noise points,we introduced the modified Levenberg-Marquardt method and the rectangular window method to establish the singular value decomposition coupled with modified Levenberg-Marquardt(SVD-mLM)method,and improved the hybrid inversion method. To verify the improved method,we designed a five-story single-span steel frame model for shaking table test. The real noise-contaminated dynamic responses are recorded,and the parameter identification and ground motion inversion are carried out for the model by the improved composite inversion algorithm. The research results of the shaking table test show that the results of structural parameter identification and the inputting inversion are receivable,and verify that the SVD-mLM method,the rectangular window method and the improved hybrid inversion method for identifying the structural physical parameters and inversing the ground motion in time domain are feasibility and effective.