Abstract:

Industries are responsible for much of the world’s energy consumption and pollutant emissions. There is a growing awareness of society on these issues and the emergence of increasingly rigid laws. Thereby, industries seek ways to improve their industrial processes. This work addresses the unrelated parallel machine scheduling problem with setup times, seeking to minimize the makespan and the total energy consumption. This version of the problem that minimizes the total energy consumption has been recently introduced in the literature and until then only exact methods were employed for its resolution. In this work, we propose an adaptive heuristic algorithm that uses learning techniques to improve the search process. In the computational experiments, we use instances from the literature and the results of an exact method for validation. The results were examined by the hypervolume indicator and graphical analysis. In this experiment environment, the results suggest that the proposed algorithm obtained a good convergence to the Pareto front and showed that it has practical applicability.