Abstract
In this dissertation, we propose solving methods for the job shop scheduling problem with sequence-dependent setup times, considering four different objective functions. The formal properties of the problem are studied. Two new neighborhood structures are proposed, with their respective feasibility and non-improving conditions, as well as an algorithm for fast neighbors' cost estimation. These structures are embedded in local search procedures, and their hybridization with a genetic algorithm is also studied. The experimental results show that the proposed methods obtain excellent results, improving in many cases the state of the art.