An enhanced approach for shape optimization using an adaptive algorithm

This paper details the application of evolutionary strategies on shape optimization problems. Evolutionary algorithms are still rather unpopular methods due to their lack of mathematical foundation. In spite of this, in complex problems, like multi-modal and multi-objective problems, heuristic techniques are able to outperform conventional optimization methods. We give an insight into the framework of evolutionary algorithms and demonstrate their advantages on examples where gradient-based methods proved to be second-best. As is well known, the drawback of these methods is a large number of function evaluations causing huge computational overhead. We show that heuristic techniques yield satisfiable results within reasonable limits of the number of function evaluations. Numerical examples are given.