The efficiency of metaheuristic algorithms has a direct relationship with their parameters setting, so that the incorrect selection of completely effective algorithmic parameters could make them inefficient. In this research, the combination of Taguchi approach and the Data Envelopment Analysis (DEA) method are applied to enhance the efficiency of the genetic algorithm to solve the Reentrant Permutation Flow Shop (RPFS) problem. Various scenarios are formed to select genetic algorithm’s operators for units under evaluation. First, using the Taguchi method for each unit, the optimal parameters are specified with the goal of minimizing the objective function (maximum tardiness). Then the effective units are determined and ranked in order to specify the best operators of the algorithm according to the optimal objective function in the shortest possible time. This research can be used as a method for setting the parameters of other evolutionary and metaheuristic algorithms in order to avoid the disadvantages of the trial and error methods.

Keywords: Scheduling, Reentrant Permutation Flow Shop, Setting Genetic Algorithm Parameters and Operators, Data Envelopment Analysis, Taguchi Experimental Design, Andersen-Petersen Ranking Model.

Full-Text [PDF 996 kb]   (905 Downloads)