Bayesian analysis of continuous-time Markov chain (CTMC) for modelling cascading failure in reliability
Rakesh Ranjan, Banaras Hindu University.
23rd of October 2025
Abstract:
Authors: Rakesh Ranjan and Simon P Wilson
This article considers Bayesian modelling and inference of cascading failure in systems with exchangeable components. A continuous-time Markov chain is developed that allows for component repair and system failure to reflect the system structure. It also models the nature of cascading failure, where system failure arises due to a sudden cascade of component failures. Special cases of series and parallel systems are also analysed. Inference is implemented both by deriving the posterior mode through Newton's method and the EM algorithm, as well as sampling-based approaches for the posterior distribution with MCMC. Numerical illustrations are provided.