ON IDENTIFIABILITY IN THE AUTOPSY MODEL OF RELIABILITY SYSTEMS OF LIFEMONITORED AND CONDITIONALLY LIFEMONITORED COMPONENTS

The autopsy data of a coherent system is the pair (T, I), where T is t he time to the failure of the system, and I is the set of components f ailed by that time. A basic question is under which conditions the dis tribution of the autopsy data determines the joint life distribution o f the components. This problem has been studied under various distribu tional and structural assumptions (always under the assumption of inde pendent component life distributions) by Meilijson (1981), Nowik (1990 ) and Antoine et al. (1993). In this paper we study the same problem i n the situation when the autopsy data is supplemented with data from l ifemonitoring and conditional lifemonitoring of some components, as in troduced by Meilijson (1994).