A risk-based approach applied to system engineering projects: a new learning based multi-criteria decision support tool based on an ant colony algorithm

Majda Lachhab (1) , Cédrick Béler (1) , Thierry Coudert (1)

1 LGP - Laboratoire Génie de Production

This article proposes a multi-criteria decision support tool fully integrated within system engineering and project management processes that allows decision makers to select an optimal scenario of a project. A model based on an oriented graph includes all the alternative choices of a new system’s conception and realization. These choices take into account the risks inherent to perform project tasks in terms of cost and duration. The model of the graph is constructed by considering all the collaborative decisions of the different actors involved in the project. This decision support tool is based on an Ant Colony Algorithm (ACO) for its ability to provide optimal solutions in a reasonable amount of time. The model developed is a multi-objective new ant colony algorithm based on an innovative learning mechanism (named MONACO) that allows ants to learn from their previous choices in order to influence the future ones. The objectives to be minimized are the total cost of the project, its global duration and the risk associated with these criteria. The risk is modeled as an uncertainty related to the increase of the nominal values of cost and duration. The optimization tool is a part of an integrated and more global process, based on industrial standards (the System Engineering process and the Project Management one) that are widely known and used in companies.

Mots clés: Project management; System engineering; Uncertainty; Risk; Decision support; Learning

Domaines: Intelligence artificielle [cs.AI]

Majda Lachhab, Cédrick Béler, Thierry Coudert. A risk-based approach applied to system engineering projects: a new learning based multi-criteria decision support tool based on an ant colony algorithm. Engineering Applications of Artificial Intelligence, 2018, vol. 72, pp. 310-326. ⟨10.1016/j.engappai.2018.04.001⟩. ⟨hal-01826286⟩