Comparison of NSGA-II and RNSGA-II Multiobjective Genetic Algorithms for Sequencing Optimization in Unrelated Parallel Machines

This study presents an approach to validate and compare the NSGA-II and RNSGA-II multiobjective genetic algorithms (GA) applied to sequencing in unrelated parallel machines. The goal is to simultaneously optimize makespan and total energy consumption (TEC) by generating the Pareto boundary and comparing the solutions obtained between the GAs and with the single-purpose ones of the GNU Linear Programming Kit (GLPK). The results demonstrate that NSGA-II consistently outperforms RNSGA-II in solution diversity and efficiency in minimizing objectives. The NSGA-II presented 88.33% of the solutions not mastered. In addition, NSGA-II had better hypervolume values in 70% of the instances analyzed. In terms of makespan, NSGA-II achieved better results in 92% of instances and was superior in all TEC assessments. The execution time of NSGA-II was also consistently shorter compared to RNSGA-II.