Initialization in genetic algorithms for constraint satisfaction problems

Abstract

In this paper we propose a strategy to incorporate heuristic knowledge into the initial population of a Genetic Algorithm to solve Job Shop Scheduling problems. This is a generalization of strategy we proposed in a previous work. The experimental results reported confirm that the new strategy improves the former one. In particular, a higher diversity in achieved among the heuristic individuals, and at the same time the mean fitness is improved. Moreover, these improvements translate into a better convergence of the GA. © Springer-Verlag Berlin Heidelberg 2001.