An efficient algorithm for a certain class of robust optimization problems

Abstract

Purpose

The purpose of this paper is to present an efficient method for the numerical treatment of robust optimization problems with absolute reliability constraints.

Design/methodology/approach

Optimization with anti-optimization based on response surface techniques; polynomial chaos for approximation of the stochastic objective function.

Findings

The number of function calls is comparable to that of the corresponding deterministic problem. Thus, the method is well suited for complex technical systems. The performance of the method is demonstrated on an optimal design problem for turbochargers.

Originality/value

The highlights of this paper are: algorithms for robust and deterministic problems show comparable complexity; no derivatives required; good convergence properties because of special set up of optimization problem; application in complex industrial examples.

Citation

Paffrath, M. and Wever, U. (2014), "An efficient algorithm for a certain class of robust optimization problems", COMPEL - The international journal for computation and mathematics in electrical and electronic engineering, Vol. 33 No. 4, pp. 1208-1219. https://doi.org/10.1108/COMPEL-11-2012-0320