Planificación local de tareas con sistemas basados en reglas borrosas. Aplicación a la detección de defectos en tableros de fibra

Resumo

Tasks scheduling in Grid systems challenges new research capability, dueto the high heterogeneity and dynamism of the resources making up Grid systems. In this thesis, two schedulers considering the state and the heterogeneity of Grid resources to carry out the execution of a set of tasks are proposed. 8oth schedulers have been verified in several heterogeneous scenarios, demonstrating its applicability in Grid. In arder to verify the fuzzy rules in Grid scheduling, a Fuzzy Local Scheduler is developed over a Desktop Grid system. The goal is to detect defects in fiberboard melamine paper, using for such purpose the image analysis. The efficiency of the Local Scheduler is analyzed in terms of the time reduction for defect detection as well as the rate of detected defects.