Un algoritmo multi-objetivo NSGA-II para la programación de los pasos de laminación en caliente de aceros

The problem of Scheduling Hot Rolling Pass of Steels is very complex because it has large spaces search for solutions. The aim of this work was the Multi-Objective Optimization (MOO) of hot rolling steel sheets in a finishing train of six steps, implementing the second generation of Nondominated Sorting Genetic Algorithm (NSGA-II). It was used two different crossover (AG1 y AG2) operators and two objective functions. These were the total time of rolling and bending of the work rolls. Two kinds of computational experiments were performed: a) Search for the best solution, b) benchmarking of both NSGA-II configurations. For each experiment was obtained 3000 solutions. Rolling time, separation roll force, and exit thickness for each rolling stand are calculated. A flow stress constitutive model for roll force was used, valid for any low, medium carbon and Nb, Ti, V microalloyed steels. The solutions are reported, choosing some of them between 100 based on the lower computing time. The 100 solutions were generated in times lower than 70 s. AG1 had lower computing times AG2. The NSGA-II implementation minimized both the rolling time as crown of the work rolls, at low computational costs and ensuring that the flatness of strip will always be lower than international standards.