Modeling the performance of multiple-phased systems based on Bayesian network

Multiple-phased systems are defined as systems which operate throughout several sequential and distinct periods of time, during which the modes and consequences of failure differ from one another. In order to complete a task of multiple-phased system a different subset of the system´s capabilities need to function. For a multiple-phased system, its component or subsystem failures may occur at any time during the mission, but not affect the system performance until the phase in which their condition is critical. This may mean that the transition from one phase to the next is a critical event that leads to phase and mission failure, with the root cause being a component failure in a previous phase. It is shown that modeling the reliability of multiple-phased system and phased missions scenario is difficult. However, multiple-phased system is a common scenario in engineering. In order to obtain the mission reliability of multiple-phased systems, the Bayesian network reliability model is built. Firstly, each phase is represented by a Bayesian network, named phase Bayesian network. Then all phases Bayesian networks are combined by connecting the root nodes that represent the same component but belong to different phases. And we connect the leaf nodes of phase Bayesian network with a new node representing the whole multiple-phased system mission. Performance of the multiple-phased system is modeled by discrete time Bayesian network. The model features are described and the approach demonstrated.