Decomposition Method for Mode Shape Identification Using Measured Data

Abstract

A mode shape identification procedure which yields mass-weighted orthogonal modes is proposed for application to a vibration test. The orthogonalization is accomplished by eliminating errors existing in the combination of measured forced response equations and an assumed mass matrix. A matrix decomposition technique is utilized for extracting the error coefficients, by which the modification of measured data is performed subject to the theoretical constraints of the forced response equation. The procedure presented herein is formulated for a base excitation test and demonstrated numerically in evaluating the mode participation factors of a component model using a simple plate model. The identified parameters are in good agreement with the exact values, and also, the identified mode shapes satisfy the orthogonality requirement. A comparison with Targoff's method and an evaluation using simulated measurement data with random errors are also presented.