In Proceedings of the 5th World Congress of Structural and Multidisciplinary Optimization, Lido di Jesolo, Italy
Authors: Gerhard Venter, Raphael T. Haftka, and Jaroslaw Sobieszczanski-SobieskiPublication Date: May 19-23, 2003
Abstract:
The paper makes use of design points evaluated during a conventional genetic algorithm or particle swarm optimization to perform robust optimization that accounts for uncertainty in the design variables. The design points obtained from the two algorithms are used to construct response surface approximations that provide an estimate of the sensitivity to variation in the design variables. As an example, the robust optimization of a composite laminate is considered. The proposed approach allows the inclusion of the effects of uncertainty at little or no additional computational cost. The paper shows that the response surface approximations can be easily updated without keeping track of the entire history of the population. In addition a simple fading procedure is proposed that can effectively emphasize more recent information when constructing the approximation. The paper includes results from both a genetic algorithm and a particle swarm optimization algorithm and a comparison between the two algorithms.