Evaluating the Performance of Distributed Approaches for Modal Identification
In this paper two modal identification approaches appropriate for use in a distributed computing environment are applied to a full-scale, complex structure. The Natural Excitation Technique (NExT) is used in conjunction with a condensed Eigensystem Realization Algorithm (ERA), and the Frequency Domain Decomposition with Peak-Picking (FDD-PP) are both applied to sensor data acquired from a 57.5-ft, 10 bay highway sign truss structure. Monte-Carlo simulations are performed on a numerical example to investigate the statistical properties and sensitivity to noise of the two distributed algorithms. Experimental results are provided and discussed.