Deriving prognostic continuous time Bayesian networks from D-matrices

Perreault, Logan; Thornton, Monica; Strasser, Shane; Sheppard, John W.

doi:10.1109/autest.2015.7356482

Cited by 8 publications

(10 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A reliability analysis based on CTBN was developed by Boudali and Dugan (2006). More recently, a prognostics method based on CTBN was introduced to account for PHM sensors in a system (Perreault, Thornton, Strasser, & Sheppard, 2015). This method was applied to a system in order to predict faults and act on them to prevent system failure.…”

Section: Phm-potential Methodsmentioning

confidence: 99%

Prognostic systems representation in a function-based Bayesian model during engineering design

L'Her

Bossuyt²,

O’Halloran

2020

IJPHM

View full text Add to dashboard Cite

Prognostics and Health Management (PHM) systems are usually only considered and set up in the late stage of design or even during the system’s lifetime, after the major design decision have been made. However, considering the PHM system’s impact on the system failure probabilities can benefit the system design early on and subsequently reduce costs. The identification of failure paths in the early phases of engineering design can guide the designer toward a safer, more reliable and cost-efficient design. Several functional failure modeling methods have been developed recently. One of their advantages is to allow for risk assessment in the early stages of the design. Risk and reliability functional failure analysis methods currently developed do not explicitly model the PHM equipment used to identify and prevent potential system failures. This paper proposes a framework to optimize prognostic systems selection and positioning during the early stages of a complex system design. A Bayesian network, incorporating the PHM systems, is used to analyze the functional model and failure propagation. The algorithm developed within the proposed framework returns the optimized placement of PHM hardware in the complex system, allowing the designer to evaluate the need for system improvement. A design tool was developed to automatically apply the proposed method. A generic pressurized water nuclear reactor primary coolant loop system is used to present a case study illustrating the proposed framework. The results obtained for this particular case study demonstrate the promise of the method introduced in this paper. The case study notably exhibits how the proposed framework can be used to support engineering design teams in making better informed decisions early in the design phase.

show abstract

Section: Phm-potential Methodsmentioning

confidence: 99%

Prognostic systems representation in a function-based Bayesian model during engineering design

L'Her

Bossuyt²,

O’Halloran

2020

IJPHM

View full text Add to dashboard Cite

show abstract

“…Each fault has exactly one intensity matrix as it has no parents; however, for tests, an intensity matrix is needed for each state of each fault it detects, giving an upper bound on time complexity of O(c m ) where c is the maximum number of states of the faults the test detects, and m is number of faults that the test detects. This demonstrates the importance of unit tests or tests that detect faults on a small number of components, (Perreault, Thornton, Strasser, & Sheppard, 2015). An example conversion of the D-matrix from Table 3 to CTBN is shown in Figure 2.…”

Section: Mapping a D-matrix To A Ctbnmentioning

confidence: 96%

“…Creating a CTBN from a D-matrix is straightforward and capitalizes on the performance gains of the independent graphical model representation. Details of the conversion process can be found in (Perreault, Thornton, Strasser, & Sheppard, 2015). Other graphical model representations have similar performance gains, but since the fault events are also dependent on time with no time step appropriate for all faults, we believe the CTBN with its associated conditional CTMPs to be a more accurate representation.…”

Section: Mapping a D-matrix To A Ctbnmentioning

confidence: 99%

“…If a failure occurs, then the repair rate, µ, parameterizes an exponential density indicating the probability at which time the failure will be fixed. The repair rate can be omitted if the failure is unrepairable (Perreault, Thornton, & Sheppard, 2016a). An example conversion of the fault tree in Figure 1 is shown in Figure 3.…”

Section: Mapping a Fault Tree To A Ctbnmentioning

confidence: 99%

See 1 more Smart Citation

Continuous Time Bayesian Networks in Prognosis and Health Management of Centrifugal Pumps

et al. 2019

View full text Add to dashboard Cite

This paper presents a novel method for performing risk-based prognosis and health management (rPHM) on centrifugal pumps. We present the rPHM framework and apply common modeling tools used in reliability and testability analysis---dependency (D) matrices and fault tree analysis---as a basis for constructing an underlying predictive model. We then introduce the mathematics of the Continuous Time Bayesian Network (CTBN), which is a probabilistic graphical model based on a factored Markov process that is designed to capture system evolution through time, and we explain how to apply a CTBN derived from D-matrices and fault trees to consider the impact of a set of faults common to centrifugal pumps on emerging hazards in the pump system. We demonstrate the utility of using CTBNs for rPHM analysis with two experiments showing the descriptive power of our modeling approach.

show abstract

“…NoisyOR functions were used to determine Boolean variables as we preferred to quantify the effect of each causal factor on its parent node independently of considering all possible combinations of states of the other parents. The NoisyOR function simplifies the elicitation of complex conditional probability tables and soothes the presumption that a factor can be reported as a "True" state only when a parent is also in the "True" status ( Kyburg and Pearl, 1991 ; Perreault et al, 2016 ). This is demonstrated by introducing the 'leak' factor which suggests that there are other unknown parent variables (nodes).…”

Section: Bayesian Belief Network (Bbn)mentioning

confidence: 99%

Developing a resilience assessment model for critical infrastructures: The case of port in tackling the impacts posed by the Covid-19 pandemic

Panahi

Gargari

Lau

et al. 2022

Ocean & Coastal Management

View full text Add to dashboard Cite

Deriving prognostic continuous time Bayesian networks from D-matrices

Cited by 8 publications

References 13 publications

Prognostic systems representation in a function-based Bayesian model during engineering design

Prognostic systems representation in a function-based Bayesian model during engineering design

Continuous Time Bayesian Networks in Prognosis and Health Management of Centrifugal Pumps

Developing a resilience assessment model for critical infrastructures: The case of port in tackling the impacts posed by the Covid-19 pandemic

Contact Info

Product

Resources

About