Issue |
Int. J. Simul. Multidisci. Des. Optim.
Volume 2, Number 1, January 2008
|
|
---|---|---|
Page(s) | 49 - 55 | |
DOI | https://doi.org/10.1051/smdo:2008006 | |
Published online | 16 May 2008 |
An efficient global optimization algorithm based on augmented radial basis function
1
Centre of Numerical Simulation for Engineering, Beijing University of Technology, F100022 Beijing, P.R. China
2
Laboratory of Mechanics, Materials & Structures, University of Reims Champagne-Ardenne, 51687 Reims, France
Corresponding authors: ysui@bjut.edu.cn lsp102@emails.bjut.edu.cn yq.guo@univ-reims.fr
Received:
29
August
2007
Accepted:
18
February
2008
In the structural optimization, the accuracy of approximation for the established mathematical model will directly affect the solution efficiency, even the convergence. The global optimization model based on the augmented Gaussian radial basis function h as a high approximation accuracy, but the solution efficiency will not be increased without a matched optimization algorithm. In this paper, we adopt the information at the interpolating points in large extent and the augmented Gaussian radial basis function to construct the approximate mathematical model. Using the explicit derivatives of the model for the sensitivities and sequential quadratic programming (SQP) algorithm for the optimization solving, an efficient algorithm of global optimization is proposed. It is simple to be realized and converges quickly. Two examples will illustrate the stability and efficiency of the present algorithm.
Key words: Structural optimization / approximate model / Gaussian radial basis function / global optimization algorithm
© ASMDO, EDP Sciences, 2008
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.