Improving heuristic estimations with constraint propagation in searching for optimal schedules

We face the Job Shop Scheduling Problem by means of branch and bound and A ∗ search. Our main contribution is a new method, based on constraint propagation rules, that allows improving the heuristic estimations. We report results from an experimental study across conventional instances with differen...

Descripción completa

Detalles Bibliográficos
Autores: Mencía Cascallana, Carlos|||0000-0001-7361-5709, Sierra Sánchez, María Rita|||0000-0002-4884-5243, Varela Arias, José Ramiro|||0000-0002-1610-1792
Tipo de recurso: capítulo de libro
Fecha de publicación:2009
País:España
Institución:Universidad de Oviedo (UNIOVI)
Repositorio:RUO. Repositorio Institucional de la Universidad de Oviedo
Idioma:inglés
OAI Identifier:oai:digibuo.uniovi.es:10651/34022
Acceso en línea:http://hdl.handle.net/10651/34022
Access Level:acceso abierto
Palabra clave:Job shop scheduling
Heuristic search
A* algorithm
Branch and bound
Constraint propagation
Descripción
Sumario:We face the Job Shop Scheduling Problem by means of branch and bound and A ∗ search. Our main contribution is a new method, based on constraint propagation rules, that allows improving the heuristic estimations. We report results from an experimental study across conventional instances with different sizes showing that A ∗ takes profit from the improved estimations. Both algorithms can reach optimal solutions for medium size instances and, in this case, the branch and bound algorithm is better than A ∗ . However, for very large instances that remain unsolved in both cases, A ∗ returns much better lower bounds due to the improved estimation