Application of principal component analysis methods to experimental structural dynamics

Abstract

Principal Component Analysis (PCA) methods have been variously developed and applied within the experimental modal community for some time based upon the underlying linear/superposition nature of structural dynamics. While historically the use of these techniques has been restricted to the areas of model order determination (Complex Mode Indicator Function [CMIF]), enhanced frequency response function estimation, and parameter identification, increasingly they are being applied to the areas of test/model validation, experimental model correlation/repeatability and experimental/structural model comparison. With the increasing volume of data being collected today, techniques which provide effective extraction of the significant data features for quick, easy comparison are essential. This paper explores the general development and application of PCA to experimental modal analysis and its ability to provide the analyst with an effective global trend visualization tool. As examples, the results from a laboratory test structure (circular plate), a civil infrastructure (bridge), and a comparative study (automotive) are presented. ©2010 Society for Experimental Mechanics Inc.