Abstract

This paper introduces a variant of the metaheuristic Simulated Annealing, oriented to solve multiobjective optimization problems. This technique is called MultiObjective Simulated Annealing with Random Trajectory Search (MOSARTS). This technique incorporates short an long term memory concepts to Simulated Annealing in order to balance the search effort among all the objectives involved in the problem. The algorithm was tested against three different techniques on a real life parallel machine scheduling problem, composed of 24 jobs and two identical machines. This problem represents a real life case study of the local sawmill industry. The results showed that MOSARTS behaved much better than the other methods utilized, because found better solutions in terms of dominance and frontier dispersion.