A PARAMETER ESTIMATION METHODOLOGY USING DYNAMIC RESPONSE TIME SERIES

A novel time domain soft parameter estimation methodology for a base-excited multi-degree-of- freedom (MDOF) structure using response and base excitation acceleration time series is proposed in this paper. For the purpose of parameter estimation, a reference structure is assumed. Without eigen- values extraction from measurements, a parametric evaluation neural network (PENN) is constructed to facilitate the inter-story stiffness identification process. For the training of PENN, a number of associated structures with structure parameters within a certain interested region are constructed. The root mean square (RMS) values of the displacement difference between each associated structure and the object structure is treated as input tothe PENN, and the corresponding inter-storey stiffness of the associated structure is the output. The performance of the proposed methodology and the sensibility of the evaluation index of RMS values are examined with simulated measurements of a 5-story frame structure under base excitation. Results show that the proposed methodology may be an applicable method for structural damage detection.
Dynamic measurement stiffness identification neural network base excitation
B. Xu
College of Civil Engineering, Hunan University, Changsha, China
国际会议
上海
英文
1500-1508
2007-11-28(万方平台首次上网日期,不代表论文的发表时间)