A Hybrid Evolutionary Algorithm for CSP

Abstract

In this paper, we present a new framework for combining complete and incomplete methods in order to solve constraint satisfaction problems. This algorithm scheme uses constraint propagation techniques and local search heuristics embedded in an evolutionary computation context. The uniformity of the involved structures provides an harmonious interaction between the different implemented methods, and also allows to take advantage of the respective methods. Furthermore, the great flexibility of this model allows us to foresee various extensions. We emphasize the interest of our approach on some examples which are solved by means of an implementation. © Springer-Verlag 2004.