Structural damage identification using co-evolution and frequency response functions

A new damage identification strategy is presented in which damage scenarios and optimal tests are searched simultaneously. The proposed strategy, called Estimation-Exploration Algorithm (EEA), is based on co-evolutionary principles. Co-evolution is a biological process where populations of interacting individuals challenge each other in an ongoing cycle of adaptation. In EEA, a population of damage hypotheses evolves to predict physical tests that have been performed on a structure, while tests evolve to create discrepancy among current damage hypotheses that can explain the observed data. This co-evolutionary approach leads to physical experiments that carry optimal information and results in a fewer number of tests needed for the correct identification of the current damage state of a structure. EEA was introduced by the authors in the context of static testing and is extended in this work to steady-state dynamics. In this context, changes in frequency response functions are used to locate and quantify damage, while structural tests are defined by the location of excitation forces and sensors position. This work shows that EEA is a feasible methodology to alleviate the ill-posedness of inverse damage detection problems by providing an intelligent strategy for selecting tests that maximize information. Copyright © 2007 John Wiley & Sons, Ltd.

Duke Scholars

Author Wilkins Aquino Thomas Lord Department of Mechanical Engineering and Materia ...