Abstract:

This work deals with the unrelated parallel machine scheduling problem with sequencedependent setup times, with the objective of minimizing the makespan. This problem appears in different industries as textile and chemicals, and its resolution is challenging due to its complexity. It was proposed for its resolution an algorithm based on the Adaptive Large Neighborhood Search method with learning automata to adapt the probabilities of using removal and insertion methods. One of the main insertion methods is inspired on the Hungarian algorithm, that is able to solve subproblems optimally. In the computational experiments were used a subset of instances of literature. The results were compared with two other algorithms of literature. For the experimental context the results suggest that the proposed algorithm found the best results in 93% of cases. This can indicate the efficiency of the proposed algorithm to the established conditions.