Modeling Supply Chain Diagnostics with Fuzzy Dynamic Bayesian Networks

Bayesian networks have been widely used as knowledge bases under uncertainty. However, in previous works, the uncertainty measure in Bayesian networks are usually probability distributions for crisp variables, which restricts the practical usefulness when incomplete knowledge or linguistic vagueness is involved in reasoning systems. This study develops a fuzzy dynamic Bayesian network (FDBN) in which fuzzy variables as well as crisp variables are considered. The proposed fuzzy dynamic Bayesian network is applied to supply chain modeling and reasoning. The simulation algorithms are designed to answer various diagnostic queries from supply chains. Significance: This work extends conventional Bayesian networks into fuzzy dynamic Bayesian networks for supply chain diagnostics. With an illustrative case, this study demonstrates how supply chain inefficiency can be modeled and diagnosed with FDBN.