Fast belief estimation in evidence network models

This paper introduces a novel approach to model complex engineering systems in the form of a network of interconnected nodes. Each node is associated to a value and an evidence measure associated to that value. The Belief and Plausibility associated to the total value of the network is then estimated with a fast decomposition technique that allows for several order of magnitude reduction in computational time under some assumptions on the properties of the network. The modelling approach and associated Belief estimation technique are proposed for the optimisation of complex engineering systems under epistemic uncertainty. The methodology is applied to the preliminary design of a small satellite where some quantities are affected by an epistemic uncertainty. In addition, the paper describes a surrogate method that provides a faster evaluation of the belief curve.