Diagnosability improvement of dynamic clustering through automatic learning of discrete event models

Nathalie A Barbosa (1) , Louise Travé-Massuyès (1) , Victor H Grisales (2)

1 LAAS-DISCO - Équipe DIagnostic, Supervision et COnduite
2 UNAL - Universidad Nacional de Colombia [Bogotà]

This paper deals with the problem of improving data-based diagnosis of continuous systems taking advantage of the system control information represented as discrete event dynamics. The approach starts from dynamic clustering results and, combining the information about operational modes, automatically generates a discrete event system that improves clustering results interpretability for decision-making purposes and enhances fault detection capabilities by the inclusion of event related dynamics. The generated timed discrete event system is adaptive thanks to the dynamic nature of the clusterer from which it was learned, namely DyClee. The timed discrete event system brings valuable temporal information to distinguish behaviors that are non-diagnosable based solely on the clustering itself.

Mots clés: Machine learning; Situation assessment; Clustering; DES; Diagnosability

Domaines: Intelligence artificielle [cs.AI]

Nathalie A Barbosa, Louise Travé-Massuyès, Victor H Grisales. Diagnosability improvement of dynamic clustering through automatic learning of discrete event models. IFAC-PapersOnLine, 2017, 50 (1), pp.1037-1042. ⟨10.1016/j.ifacol.2017.08.214⟩. ⟨hal-02004430⟩