Mining gene structures to inject artificial chromosomes for genetic algorithm in single machine scheduling problems

In this paper, a genetic algorithm with injecting artificial chromosomes is developed to solve the single machine scheduling problems. Artificial chromosomes are generated according to a probability matrix which is transformed from the dominance matrix by mining the gene structure of an elite base. A roulette wheel selection method is applied to generate an artificial chromosome by assigning genes onto each position according to the probability matrix. The higher the probability is, the more possible that the job will show up in that particular position. By injecting these artificial chromosomes, the genetic algorithm will speed up the convergence of the evolutionary processes. Intensive experimental results indicate that proposed algorithm is very encouraging and it can improve the solution quality significantly. (C) 2007 Elsevier B.V. All rights reserved.