Sistema para la optimización del proceso de secuenciación de reportes con múltiples restricciones
Fecha
2016-05-18
Autores
Coto Palacio, Jessica
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Central “Marta Abreu” de Las Villas
Resumen
Los problemas de secuenciación de tareas aparecen de forma frecuente en disímiles situaciones de la vida real, siempre que sea necesario asignar recursos a la realización de tareas en espacios de tiempo, optimizando una o más funciones objetivo. En dependencia del problema a resolver dichas tareas pueden tomar diferentes formas, y los objetivos también pueden variar. En esta investigación se aborda la secuenciación de tareas en ambientes de manufactura, donde es necesario secuenciar los reportes solicitados por los clientes en un conjunto de máquinas con restricciones de capacidad. Adicionalmente aparecen otra serie de limitaciones que es necesario tener en cuenta al construir una solución factible, lo cual hace que el sistema utilizado actualmente no resuelva de manera eficiente el problema de secuenciación de reportes, teniendo en cuenta todas las restricciones que presenta la empresa. Para resolver el problema en cuestión se propone un algoritmo general, el cual inicialmente distribuye la capacidad total del sistema entre los recursos existentes, teniendo en cuenta las capacidades que presentan los mismos, siguiendo una de cuatro posibles alternativas de selección. Posteriormente cada recurso selecciona en qué orden procesará los reportes que tiene asignados. El estudio experimental desarrollado muestra que las cuatro alternativas implementadas permiten obtener soluciones factibles para el problema de secuenciación de reportes planteado, siendo la alternativa basada en Q-Learning la que mejores resultados obtiene.
Scheduling problems appear on a regular basis in many real life situations, whenever it is necessary to allocate resources to perform tasks, optimizing one or more objective functions. Depending on the problem being solved, these tasks can take different forms, and the objectives can also vary. This research addresses scheduling in manufacturing environments, where it is necessary to sequence the reports requested by customers in a set of machines with capacity constraints. Additionally, there is another set of limitations that must be taken into account when constructing a feasible solution, which means that the system currently used does not efficiently solve the report scheduling problem, taking into account all the constraints of the company. To solve this problem, a general algorithm is proposed, which initially distributes the total capacity of the system among the existing resources, taking into account the capacity presented by teach hem, and following one of four possible selection alternatives. Subsequently, each resource decides in which order it will process the reports assigned to it. The experimental study performed shows that the four alternatives implemented allow to obtain feasible solutions for the report scheduling problem, being the alternative based on Q-Learning the one that obtains the best results.
Scheduling problems appear on a regular basis in many real life situations, whenever it is necessary to allocate resources to perform tasks, optimizing one or more objective functions. Depending on the problem being solved, these tasks can take different forms, and the objectives can also vary. This research addresses scheduling in manufacturing environments, where it is necessary to sequence the reports requested by customers in a set of machines with capacity constraints. Additionally, there is another set of limitations that must be taken into account when constructing a feasible solution, which means that the system currently used does not efficiently solve the report scheduling problem, taking into account all the constraints of the company. To solve this problem, a general algorithm is proposed, which initially distributes the total capacity of the system among the existing resources, taking into account the capacity presented by teach hem, and following one of four possible selection alternatives. Subsequently, each resource decides in which order it will process the reports assigned to it. The experimental study performed shows that the four alternatives implemented allow to obtain feasible solutions for the report scheduling problem, being the alternative based on Q-Learning the one that obtains the best results.
Descripción
Palabras clave
Secuenciación de Reportes, Optimización, Q-Learning