Nonlinear digital system identification applied to vibrating structures
Abstract
The analysis of a specific nonlinear system depends on having an accurate model of the system and an analysis technique that allows exploration of the possible types of behavior. In rotating machinery, the excitation is typically harmonic and the coupling of this excitation with particular system resonances is of interest, as are the responses that result from this coupling. Prior to analysis, accurate models must be determined from measurements. A discrete modeling system identifications technique (NARMAX modeling) is studied and modified to understand and improve its performance in the presence of measurement artifacts such as anti-aliasing filtering, quantization noise and sampling. Necessary excitations for identifications are explored through simulation prior to applying the techniques to two experimental systems: a thin beam oscillating between two magnets and a thin tensioned plate with nearly coincidental natural frequencies. A technique, analogous to continuous-time averaging of differential equations with sinusoidal excitations, is developed to analyze these discrete-time difference equation models. Regions of interesting behavior can be identified as a function of model parameters and excitation parameter variation, thus providing a guide to experimentalists to approach regions where particular nonlinear behavior (i.e., jumps, amplitude modulations, and chaos) may be observed.
Degree
Ph.D.
Advisors
Davies, Purdue University.
Subject Area
Mechanical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.