Desarrollo de técnicas de búsqueda metaheurísticas en problemas de scheduling multiobjetivo
In Artificial Intelligence field, scheduling problems are applied to multiple real environments and its main objective is to find optimized solutions by using centralized and static search techniques. However, the reality is far from this approach, where a lot of scheduling problems take place in di...
| Autor: | |
|---|---|
| Tipo de documento: | dissertação |
| Data de publicação: | 2017 |
| País: | España |
| Recursos: | Universitat Politècnica de València (UPV) |
| Repositório: | RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia |
| Idioma: | espanhol |
| OAI Identifier: | oai:riunet.upv.es:10251/86075 |
| Acesso em linha: | https://riunet.upv.es/handle/10251/86075 |
| Access Level: | Acceso aberto |
| Palavra-chave: | Scheduling Green scheduling Multiagent Production control Control de producción Multiagente LENGUAJES Y SISTEMAS INFORMATICOS Máster Universitario en Inteligencia Artificial, Reconocimiento de Formas e Imagen Digital-Màster Universitari en Intel·ligència Artificial, Reconeixement de Formes i Imatge Digital |
| Resumo: | In Artificial Intelligence field, scheduling problems are applied to multiple real environments and its main objective is to find optimized solutions by using centralized and static search techniques. However, the reality is far from this approach, where a lot of scheduling problems take place in distributed and dynamic environments. This dynamic and distributed nature of scheduling problems, along with new objectives to optimize such as energy efficiency, force us to manage them in a efficient way in order to improve computing time while it adapts to new constraints of the problem. Each available resource usually has and associated energy consumption and it is variable depending on its use time or the assigned power work. Given a context in which more and more efficient processes are needed, we need to obtain energy efficient solutions. This represents a new view in the way that we approach scheduling problems called ¿Green Scheduling¿. In this project we present a multi-agent model as a tool to solve Green Scheduling offline problems in production system with high energy consumption. In the proposed model, agents collaborate with each other in order to achieve an agreement in which we optimize a multiobjective function based on minimizing 3 factors: the total tardiness of the jobs, the total setup time and the total energy consumption that machines need to perform all the assigned jobs. Later, we add an online system integrating it in the offline system previously mentioned. This online system monitors the execution of the plan scheduled offline, waiting until an incidence appears. When an incidence occurs, the online system carries out a rescheduling of the affected machine in real time, redistributing the jobs originally assigned to it, if possible, and trying to minimize the impact of the incidence on the global costs of the whole production context. The developed systems have been evaluated on several test cases and the obtained results show the utility of the proposals. |
|---|