Improving the Multi-Restart Local Search Algorithm by Permutation Matrices and Sorted Completion Times for the Flow Shop Scheduling Problem

Abstract

Iterated local search (ILS) is a metaheuristic used successfully to solve the flow shop scheduling problem. In particular, the multi-restart ILS (MRSILS) is an easily implementable algorithm which obtains good results. In this paper, we modify the MRSILS algorithm in two ways. First, small changes in the initial solution are generated by permutation matrices in order to improve it before using the MRSILS. Second, a minor variation is made in the strategy of the MRSILS. Sorted completion times are taken to select the job to be inserted in new positions to obtain a better scheduling. The original MRSILS and both modifications are evaluated with well-known benchmark instances. The experiments show that the new modifications produce slightly better results than the original one, especially for a large number of jobs