Keywords: damage detection, genetic algorithms, dynamic parameters, finite element model, optimization, frame structures.

The damage detection problem can be considered as an optimization problem and genetic algorithms have been used to solve it, as in [1]. One of the principal difficulties to locate and quantify damage in a structure is the fact that the number and localization of the damaged elements is not known at the beginning of the optimization process. For this reason, this paper proposes a damage detection methodology that considers the above difficulty.

First, the damage is considered as a reduction in the elasticity modulus of the damaged element. This reduction is obtained by using an elasticity reduction factor, which assumes a value equal to 0 when the element is undamaged, and 1 to signify the total damage.

A multi-chromosome genetic algorithm (MGA) is used to solve the optimization problem. This type of genetic algorithm was used by Hinterding for a self-adaptation of genetic parameters [2]. A typical individual is constituted by two chromosomes. The first uses real numbers, between 0 and 1, to stand for the damage extension and the second is a binary chromosome used to locate the damaged elements, with a value equal to 1, meaning that the corresponding element is damaged. It can be observed that the number of damaged elements in different chromosomes of the same population is not necessarily the same.

The objective function was based on the difference between the experimental dynamical parameters of the damaged structure and those of an analytical model obtained by using the genetic algorithm. Here, the experimental dynamic parameters were obtained numerically from a finite element model of the damaged structure. However, noisy and incomplete measurements were considered.

The proposed methodology was applied to a beam structure and a truss structure and simple and multiple damage scenarios were analyzed. Results show that all of the damaged elements in the different damage scenarios were found. The damage extension of these elements was found with a maximum difference of 0.04. Some misidentified damaged elements were observed, but in general they presented low values of damage. The best performance of this methodology was obtained for the computation of simple damage scenarios, which were found with high accuracy.

1

C. Mares, C. Surace, "An application of genetic algorithm to identify damage in elastic structures", Journal of Sound and Vibration, 195(2), 195-215, 1996. doi:10.1006/jsvi.1996.0416

2

R. Hinterding, "Self-adaptation using multi-chromosomes", in "Proceedings of the IEEE International Conference on Evolutionary Computation", Indianapolis, United States, 87-91, 1997. doi:10.1109/ICEC.1997.592274

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £145 +P&P)