Section: New Results

Optimization under uncertainty

Participants: El-Ghazali Talbi, Raca Todosijevic, Oumayma Bahri (externel collaborators: Nahla BenAmor - Univ. Tunis, Tunisia, J. Puente, C. R. Vela, I. Gonzalez-Rodriguez - Univ. Oviedo Spain)

At the problem level, the sources of uncertainty are due to many factors such as the environment parameters of the model, the decision variables and the objective functions. Examples of such uncertainties can be the demand and travel times in vehicle routing problems, the execution time in scheduling problems, the wind or solar production in energy power systems, the price of resources in manufacturing, and the mechanical properties of a structure. Then, we need precise and efficient modeling and resolution approaches which are robust and non-sensitive to those uncertainties. The appeal of optimization under uncertainty is that its performance results remain relatively unchanged when exposed to uncertain data.

We have considered the fuzzy job shop, a job shop scheduling problem with uncertain processing times modelled as triangular fuzzy numbers. While the usual approaches to solving this problem involve adapting existing metaheuristics to the fuzzy setting, we have proposed instead to follow the framework of simheuristics from stochastic optimisation. More precisely, we integrate the simulation of possible realisations of the fuzzy problem with a genetic algorithm that solves the deterministic job shop. We test the resulting method, simGA, on a testbed of 23 benchmark instances and obtain results that suggest that this is a promising approach to solving problems with uncertainty by means of metaheuristics [38].