A framework for dynamic risk assessment with condition monitoring data and inspection data

Jinduo Xing (1) , Zhiguo Zeng (2) , Enrico Zio (3)

1 SSEC - Chaire Sciences des Systèmes et Défis Energétiques EDF/ECP/Supélec
2 LGI - Laboratoire Génie Industriel - EA 2606
3 POLIMI - Politecnico di Milano [Milan]

In this paper, a framework is proposed for integrating condition monitoring and inspection data in Dynamic Risk Assessment (DRA). Condition monitoring data are online-collected by sensors and indirectly relate to component degradation; inspection data are recorded in physical inspections that directly measure the component degradation. A Hidden Markov Gaussian Mixture Model (HM-GMM) is developed for modeling the condition monitoring data and a Bayesian network (BN) is developed to integrate the two data sources for DRA. Risk updating and prediction are exemplified on an Event Tree (ET) risk assessment model. A numerical case study and a real-world application on a Nuclear Power Plant (NPP) are performed to demonstrate the application of the proposed framework.

Domaines: Informatique [cs]

Jinduo Xing, Zhiguo Zeng, Enrico Zio. A framework for dynamic risk assessment with condition monitoring data and inspection data. Reliability Engineering and System Safety, 2019, 191, pp.106552. ⟨10.1016/j.ress.2019.106552⟩. ⟨hal-02428514⟩