Skip to Main content Skip to Navigation
New interface

Dynamic probabilistic graphical model applied to the system health diagnosis, prognosis, and the remains useful life estimation

Abstract : This thesis contributes to prognosis and health management for assessing health condition of complex systems. In the context of operational management and operational safety of systems, we propose to investigate how Dynamic Probabilistic Graphical Modelling (DPGM) can be used to diagnose the current health state of systems, prognostic the future health state, and the evolution of degradation, as well as estimate its remaining useful life based on its operating conditions. System degradation is generally unknown and requires shutting down the system to be observed. However, this is difficult or even impossible during system operation. Though, a set of observable quantities on a system or component can characterise the level of degradation and help to estimate the remaining useful life of components and systems. The DPGM provides an approach suitable for modelling the evolution of the health state of systems and components. The aim of this thesis is to transpose and capitalize on the experience of these previous works in a prognostic context on the basis of a more efficient DPGM taking into account the available knowledge on the system. We extend the classical HMM family models to the IOHMM to allow the time propagation of uncertainty to address prognostic problems. This research includes the extension of learning and inference algorithms. Variants of the HMM model are proposed to incorporate the operating environment into the prognosis. The aim of this thesis is to contribute to solving the following scientific locks: - Considering the state of health whatever the complexity of the system by a stochastic model and learning the model parameters from the available measurements on the system. - Establish a diagnosis of the state of health of the system and the prognosis of its evolution by integrating several operational conditions. - Estimate the remaining useful life of components and structured systems with series and parallel components. This is a major challenge because the prognosis of the degradation of system components makes it possible to define strategies for either control or maintenance in relation to the residual life of the system. This allows the reduction of the probability of occurrence of a shutdown due to a system malfunction either by adjusting the degradation speed to fit in with a preventive maintenance plan or by proactively planning maintenance interventions.
Document type :
Complete list of metadata
Contributor : Thèses UL Connect in order to contact the contributor
Submitted on : Monday, December 14, 2020 - 4:12:21 PM
Last modification on : Wednesday, November 3, 2021 - 7:08:47 AM
Long-term archiving on: : Monday, March 15, 2021 - 7:51:39 PM


Files produced by the author(s)


  • HAL Id : tel-03064961, version 1



Kamrul Islam Shahin. Dynamic probabilistic graphical model applied to the system health diagnosis, prognosis, and the remains useful life estimation. Automatic. Université de Lorraine, 2020. English. ⟨NNT : 2020LORR0129⟩. ⟨tel-03064961⟩



Record views


Files downloads