A Multiobjective Memetic Approach to Job-Shop Scheduling under Uncertainty

Thanh-Do Tran (1) , Inés González-Rodríguez (2) , El-Ghazali Talbi (1)

1 DOLPHIN - Parallel Cooperative Multi-criteria Optimization
2 Department of Mathematics, Statistics and Computing

In recent years, much work has been expended on addressing job-shop scheduling problems (JSP) with uncertain information. There are two primary approaches to uncertainty handling, i.e. using probability theory and possibility theory. In this work, we use the possibilistic approach to deal with JSP where uncertain processing times are modeled by triangular fuzzy numbers (TFNs). Algorithmically, this paper examines the incorporation of a local search into a multiobjective genetic approach. The incorporation results in a simple multiobjective memetic algorithm that is based on the NSGA-II and the N2 neighborhood structure for individual improvement in the Lamarckian learning procedure. An extensive experiment was conducted to confirm the superiority of the algorithm compared to both the single-objective memetic and multiobjective genetic methods.

Mots clés: job-shop scheduling; multiobjective memetic algorithm; uncertainty; robustness analysis

Domaines: Recherche opérationnelle [math.OC]

Thanh-Do Tran, Inés González-Rodríguez, El-Ghazali Talbi. A Multiobjective Memetic Approach to Job-Shop Scheduling under Uncertainty. The 5th International Conference on Metaheuristics and Nature Inspired Computing (META'14), Oct 2014, Marrakech, Morocco. ⟨hal-01110315⟩