Constrained engineering design optimization using a hybrid bi-objective evolutionary-classical methodology

Abstract

Constrained engineering design optimization problems are usually computationally expensive due to non-linearity and non convexity of the constraint functions. Penalty function methods are found to be quite popular due to their simplicity and ease of implementation, but they require an appropriate value of the penalty parameter. Bi-objective approach is one of the methods to handle constraints, in which the minimization of the constraint violation is included as an additional objective. In this paper, constrained engineering design optimization problems are solved by combining the penalty function approach with a bi-objective evolutionary approach which play complementary roles to help each other. The penalty parameter is approximated using bi-objective approach and a classical method is used for the solution of unconstrained penalized function. In this methodology, we have also eliminated the local search parameter which was needed in our previous study. © 2010 Springer-Verlag.