Availability assessment of stochastic multi-states systems based on UGF and taking into account data uncertainty
Résumé
The presented method extends availability assessment in multi state systems to take into account data uncertainty. Until now, propagation of uncertainty is evaluated only in the classical block diagrams methods or specific simulation modeling. Using the universal generating function technique (UGF), the number of states in the multi states model is reduced and the uncertainty evaluation is generalized. Markov processes describe the multiple degradation states of components (with exponential laws) which are combined with UGF to generate the system states. The reparations are also taking into account in terms of exponential laws. The data uncertainty is modeled by usual probability functions. The results show the influence of these uncertainties on the probability states, availability assessment and consequently the decision for system design or maintenance policy. The suggested method is sufficiently general to cover many types of systems and application processes.