Using dynamic Bayesian networks for prognostic modelling to inform maintenance decision making

In this paper, we consider the application of dynamic Bayesian networks to the prognostic modelling of equipment in order to better inform maintenance decision-making. We provide a brief overview of Bayesian networks and their application to reliability modelling. An example is then provided in which an equipment is considered to be in one of six states and there are two imperfect condition monitoring indicators available to provide evidence about the equipment's true state which tends to deteriorate over time. With this example, we show how the equipment's reliability decays over time in the situation where repair is not possible and then how a simple change to the model allows us to represent different maintenance policies for repairable equipment.