Comparison of solution space exploration by NSGA2 and SPEA2 for Flexible Job Shop Problem

Julien Autuori (1) , Faicel Hnaien (1) , Farouk Yalaoui (1) , Abdelaziz Hamzaoui (2) , Najib Essounbouli (2)

1 LOSI - Laboratoire d'Optimisation des Systèmes Industriels
2 CRESTIC - Centre de Recherche en Sciences et Technologies de l'Information et de la Communication - EA 3804

In this paper, we present a new metric to evaluate the meta-heuristics by studying their exploration of the solution space. We propose an efficient representation of the solution space by a one-dimensional map. To evaluate the exploration of space, relative and absolute metrics have been proposed. For the tests, we used the NSGA2 and SPEA2 on Flexible Job Shop Problem (FJSP) with the objectives to reduce the makespan (C _max ) and to produce the jobs just in time (Σ ^N _i=1 Z _i , with Z _i = T _i + E _i , T is the tardiness and E is the earliness). The results show that the NSGA2 explore more 6.78% solutions on average than SPEA2.

Mots clés: Space exploration; Measurement; Sociology; Statistics; Biological cells; Equations; Mathematical model; Multi-objectives Optimization; Flexible Job Shop Problem; Evolutionary algorithms; Solution space

Domaines: Recherche opérationnelle [math.OC]

Julien Autuori, Faicel Hnaien, Farouk Yalaoui, Abdelaziz Hamzaoui, Najib Essounbouli. Comparison of solution space exploration by NSGA2 and SPEA2 for Flexible Job Shop Problem. 2013 International Conference on Control, Decision and Information Technologies (CoDIT), May 2013, Hammamet, Tunisia. pp.750-755, ⟨10.1109/CoDIT.2013.6689636⟩. ⟨hal-02491756⟩