Heuristic rules and genetic algorithms for open shop scheduling problem

Open Shop Scheduling is a meaningful paradigm of constraint satisfaction problems. In this paper, a method combining heuristic rules and evolutionary algorithms is proposed to solve this problem. Firstly, we consider several dispatching rules taken from literature that produce semi-optimal solutions in polinomial time. From these rules we have designed probabilistic algorithms to generate heuristic chromosomes that are inserted in the initial population of a conventional genetic algorithm. The experimental results show that the initial populations generated by the proposed method exhibits a high quality, in terms of both solutions cost and diversity. This way the genetic algorithm converges to much better solutions than when it starts from a ramdom population.