An extended Bayesian belief network model of multi-agent systems for supply chain management

In this paper, we describe our on-going research on uncertainty analysis in Multi-agent Systems for Supply Chain Management (MASCM). In a MASCM, an agent consists of automation processes within a legal entity in the specific supply chain network. It conducts supply chain planning, execution and cooperation on behalf of its owner. Each day these agents have to process a large volume of data from different sources with mixed signals not to be anticipated in advance. Thus, one challenge every agent has to face in this volatile environment is to quickly identify the impact of unexpected events, and take proper adjustments in both local procedures and related cross-boundary interactions. To facilitate the study of uncertainty in the complex system of MASCM, we model agent system behaviors by abstracting its significant operational aspects as observation, propagation and update of uncertainty ifnromation. The resulting theoretical model, called an extended Bayesian Belief Network (eBBN), may serve as the basis for developing an uncertainty management component for a large-scale electronic supply chain system. We also briefly describe ways this model can be used to solve different supply chain tasks and some simulation results that demonstrate the power of this model in improving the system performance.