The Application of the Method of Least Squares in Modal Parameter Identification in the Frequency Domain

碩士 === 國立成功大學 === 航空太空工程學系碩博士班 === 101 === Under the process of modal parameter identification, modal interferences often cause problem on identifiability and affect the accuracy of system identification. Major causes of modal interferences are closer frequency, high damping ratio, non-proportional...

Full description

Bibliographic Details
Main Authors: Guo-GangTeng, 鄧國綱
Other Authors: Dar-Yun Chiang
Format: Others
Language:zh-TW
Published: 2013
Online Access:http://ndltd.ncl.edu.tw/handle/48467919382284667570
Description
Summary:碩士 === 國立成功大學 === 航空太空工程學系碩博士班 === 101 === Under the process of modal parameter identification, modal interferences often cause problem on identifiability and affect the accuracy of system identification. Major causes of modal interferences are closer frequency, high damping ratio, non-proportional damping too high…etc. In additions, the fictitious mode during identification will affect correct determinacy of structural modal. This article will investigate modal parameter identification in frequency domain and work on rational orthogonal polynomial method and least squares complex frequency domain method, which have been proposed by others earlier. According to structural dynamics theory, the frequency response function could be expressed as a rational fraction formula. Therefore, via the curve fitting process method we can obtain the coefficients of the rational fraction and then get the modal parameter. The error from curve fitting process method will cause a fictious mode. The parameter is unstable in different order number. On the contrary, the modal parameter of structural model is much stable and the system’s true modal information could be obtained. While not increasing order number, we propose a method to distinguish based on the paired feature of system characteristic values and the values of residue. Based on modal superposition and modal parameter expression of the frequency response function of non-proportional damping system, we think there will no problem on identification for non-proportional damping system. Furthermore, comparing least squares complex frequency domain method with rational orthogonal polynomial method, the structural modal parameter of the former is much more stable than that of the latter; therefore the former is better in distinguishing the fictious modes.