Forecasting design and decision paths in ship design using the ship-centric Markov decision process model

Abstract

This paper introduces a decision-making model for forecasting design and decision paths in ship design by applying eigenvector analysis to the ship-centric Markov decision process (SC-MDP) model. This paper uses the concept of composite reducible Markov processes to identify various independent design absorbing paths. An absorbing path represents the long term behavior of a temporal decision process. This method identifies the set of absorbing paths by decomposing the process into sets of inherently independent parts and thus also gives insight into the structure and relationships of the decision process. This is done by examining the set of principal eigenvectors. Two metrics are introduced. First, the set of principal eigenvectors is used to identify all independent design absorbing paths without the need for full examination of all initial conditions. Second, through the use of the Moore-Penrose pseudo-inverse, the set of principal eigenvectors is used to estimate the optimal life cycle strategy of the decision process. A case study is presented involving life cycle planning for ballast water treatment compliance of a notional container ship to show the utility of these methods and metrics.