HMM used for Component Parameters Apportionment

This paper deals with the problem of the correct allocation of the reliability failure and repair rates of simple systems composed of two elements only, once the service shop history is fully mapped and theoretical hypothesis on system parameters are available. Actually, the problem of data collection and the management of censored ones is one of the most important affecting both reliability, availability and safety analysis. Usually the design is based on a partial collection of biased and censored field data. To improve the predictions and make them more adherent to actual field system life, in this paper, the authors suggest to implement a simple homogeneous Markov model and compare the transition probabilities with the ones coming from a Hidden Markov Structure (HMS or HM Model). The results obtained allows for retrofitting the supposed system failure and repair rates. In this paper the classical Markov Model approach is presented and compared with the HMM one, on one simple and generalized example. In such framework, parameters apportionment, assumes the meaning of system parameters tuning based on the shop history outcomes to match actual system behavior.