Numerical Study on Identification of Time Varying Parameters of Vibration Systems

Yanchun Liang,Qiang Zhen,Zaishen Wang

doi:10.1155/1997/495860

Abstract

An on-line least squares algorithm has previously been successfully applied to linear vibration systems in order to identify time varying parameters. In this article the limitations of the approach and the factors affecting the identification are further examined. The existence of the nonlinear term is determined by means of the time varying characteristics of the estimated linear parameters using the linear model and the data from a time invariant nonlinear system. The identification of the time varying linear parameters is also examined in accordance with the linear model by using the data with nonlinear elements.

Highlights

Parameter identification of vibration systems has become one of the most important research areas in inverse problems of structural dynamics
The vast majority of parametric identification techniques that are used make the assumption that the vibrationary characteristics are independent of time, i.e., the structural parameters remain constant throughout modal tests
A number of parameter identification procedures based on the time invariant assumption have been developed, which can be used successfully to estimate structural parameters of time invariant systems

Summary

INTRODUCTION

Parameter identification of vibration systems has become one of the most important research areas in inverse problems of structural dynamics. A number of on-line identification techniques based on the least squares algorithm have previously been successfully applied to simulated and real data sets in order to identify time varying parameters. 70 Liang, Wang, and Zhen difference equation models and on-line versions of seven time domain system identification algorithms was examined by Cooper (1990). Three online system identification techniques were applied to real and simulated data sets by Cooper and Worden (1991a) in order to identify time varying physical parameters. Vibration tests on two mass varying systems were undertaken and the changes in the physical and modal parameters were tracked in Cooper and Worden (1991b). The estimates of the time varying linear parameters are examined in accordance with the linear model using the data with nonlinear elements. The value of L is set so that the algorithm has time to "warm up" before the error analysis is performed (Cooper, 1990)

NUMERICAL RESULTS AND DISCUSSION

X 1010 1 X 1011 1 X 1012

CONCLUSIONS