Quasi-online failure times identification of mobile heat sources in 2D geometry - LARIS - Systèmes Dynamiques et Optimisation Accéder directement au contenu
Article Dans Une Revue Journal of Process Control Année : 2024

Quasi-online failure times identification of mobile heat sources in 2D geometry

Résumé

Identifying the failure instants in thermal systems subject to 2D parabolic partial differential equations presents a significant challenge, especially when the systems involve mobile heat sources. In the context of this study, mobile heat sources are examined, along with a set of stationary sensors, while assuming known and constant-velocity trajectories for the heat sources. This research introduces a quasi-online methodology that incorporates Exponentially Weighted Moving Average (EWMA) charts for immediate failure detection. When a failure is detected via the EWMA charts, the Conjugate Gradient Method, traditionally developed for offline applications, is activated. This method is adapted to a quasi-online framework, facilitating a more rapid and precise identification of malfunctioning heat sources, the exact time of their failures, and the possibility of restoring normal operations. To assess the performance and reliability of this approach, it is compared with a Bayesian filter-based method, particularly using the Kalman filter for this purpose. Monte Carlo simulations are employed to evaluate the resilience and effectiveness of the quasi-online method, focusing on the system's sensitivity to the accuracy of sensor measurements.
Fichier principal
Vignette du fichier
JPC_QL.pdf (3.4 Mo) Télécharger le fichier
Origine Publication financée par une institution
licence

Dates et versions

hal-04470329 , version 1 (21-02-2024)

Licence

Identifiants

Citer

Mohamed Salim Bidou, L. Perez, Sylvain Verron, Laurent Autrique. Quasi-online failure times identification of mobile heat sources in 2D geometry. Journal of Process Control, 2024, 136, pp.103183. ⟨10.1016/j.jprocont.2024.103183⟩. ⟨hal-04470329⟩
16 Consultations
8 Téléchargements

Altmetric

Partager

Gmail Mastodon Facebook X LinkedIn More