Abstract

This work presents a multiobjective optimization algorithm based on a variant of the evolutionary programming field called memetic algorithms (MA). This algorithm was proposed by the authors, combines genetic evolution with local search, in the same way as traditional MA, but with the use of independent populations for each objective, as well as a mechanism to find compromise solutions (tradeoff), where local search is performed restricted by a compromise parameter. The algorithm was applied to a scheduling problem of a molding production process and compared against two multiobjective techniques available in the literature, Multiobjective Genetic Algorithm (MOGA) and Multiobjective Simulated Annealing (MOSA). The results of the proposed approach, based on the experiments performed, outperformed the benchmark techniques based on two objectives of industrial interest, such as, the total completion time (Cmax) and the total tardiness. Both objectives are minimization. This objectives have a direct impact on the process productivity as well as the capability of delivering the goods on time.