An integrated reactive tabu search for flowshop machine sequencing problems

We propose a tabu search technique for solving the m-machine, n-job scheduling problem in a flowshop production system. The objective is the minimization of the makespan. In our tabu scheme the size of the list changes in an automated way by reacting to the occurrence of cycles in the search process. The tabu also adopted an escape function to diversify the search to avoid possible strange attractors. Our reactive tabu search is integrated with a ''good'' starting solution given by a well known greedy algorithm. Our tabu is compared to one traditional tabu scheme characterized by a fixed list size. We show as our ''reactive'' tabu outdoes the other two heuristic ones. The results are obtained using a set of ''hard'' problems recently published.