Javier Cano Cancela
Universidad Rey Julian Carlos
We provide a class of models to evaluate and forecast the reliability of complex hardware/software systems, described through Reliability Block Diagrams (RBDs). We put special emphasis on Bayesian analysis and forecast, which current commercial packages provide little support on. The approach we adopt combines several conventional models in a novel way.
Blocks referring to hardware components are modelled through 'pending' Continuous Time Markov Chain models, whereas Phase-type distributions are used to modelize absorption times into failures modes.
Blocks referring to software components are modelled through a Bayesian model selection strategy, supporting on the parameter expanded likelihood and Bayes factors.
Inference and forecasting tasks with such models are described, and illustrated with an example.
A computational environment is currently being developed to fully support these tasks and is also briefly introduced here.