SYSBOOSTER, application of Data Science to surveillance of systems for detection or anticipation of dysfunctions or failures of systems

Alain Celisse (1) , Olivier Gauriau (2) , Margot Corréard (3) , Jean-François Bouin (3) , Lennart Priester (4) , Ronald Naumann (4) , Emmanuel Arbaretier (5) , Michel Kaczmarek (5) , Uwe Schmietainski (4) , Hagen Friedrich (4)

1 MODAL - MOdel for Data Analysis and Learning
2 LACODAM - Large Scale Collaborative Data Mining
3 DiaGrams Technologies [Lille]
4 Nokia Technologies
5 APSYS [Elancourt]

This paper describes a pipeline designed by INRIA, DIAGRAMS TECHNOLOGIES, APSYS, and NOKIA for addressing questions related to root-cause analysis. These different tools, enabling multivariate multiple change-point detection as well as the automatic detection of future failures, are part of the achievements of the SYSBOOSTER project. For confidentiality reasons, many data and technical information have been changed or anonymized along the paper.

Mots clés: dysfunctions; failures; data science; predictive maintenance; diagnosis; troubleshooting; prognosis; HUMS

Domaines: Sciences de l'ingénieur [physics]

Alain Celisse, Olivier Gauriau, Margot Corréard, Jean-François Bouin, Lennart Priester, et al.. SYSBOOSTER, application of Data Science to surveillance of systems for detection or anticipation of dysfunctions or failures of systems. Congrès Lambda Mu 22 « Les risques au cœur des transitions » (e-congrès) - 22e Congrès de Maîtrise des Risques et de Sûreté de Fonctionnement, Institut pour la Maîtrise des Risques, Oct 2020, Le Havre (e-congrès), France. pp.1-9. ⟨hal-03483963⟩