This paper deals with the single-machine scheduling problem with earliness and tardiness penalties. Sequence dependent setup times and distinct due windows are considered. In order to solve this problem, a three-phase heuristic approach, the so-called GTSPR, was developed. The first phase is based on GRASP and Variable Neighborhood Descent to generate an initial solution; the second phase is based on Tabu Search for solution refining, finally, Path Relinking is used as a mechanism of post-optimization. For each job sequence generated by the heuristic, an optimal timing algorithm is used to determine the completion time for each job in the job sequence. Computational experiments carried out show that GTSPR outperforms the previous algorithms found in the related literature, regarding the quality of the final solution and the average gap.
Single-Machine; GRASP; Tabu search; Variable neighborhood descent; Path relinking
