Multiobjective perspectives in unrelated parallel machine environments

This work addresses the scheduling problem in unrelated parallel machines (UPMSP) from the perspective of multiobjective optimization, considering the simultaneous minimization of makespan and total delay. The NSGA-II algorithm was used and recognized for its efficiency in obtaining non-dominated solutions in problems with conflicting objectives. A coding based on a vector of real numbers was adopted, facilitating the implementation of evolutionary operators in combinatorial problems. The experimental results demonstrated the existence of trade-offs between the objectives, with different patterns of machine occupation depending on the prioritized criterion. The heterogeneity of processing times, a characteristic of UPMSP, accentuates this dynamic, reiniforcing the importance of approaches that provide multiple alternative solutions to support decision making.