Real Time Assessment of Drinking Water Systems Using a Dynamic Bayesian Network

This paper presents a methodology for real-time estimation of water distribution system state parameters using a dynamic Bayesian network to combine current observations with knowledge of past system behavior. The dynamic Bayesian network presented here allows the flexibility to model both discrete and continuous variables and represent causal relationships that exist within the distribution system. The posterior belief state can be inferred using a compact approximation algorithm that has been shown to contain inference errors. Simulations over stochastic variables are proposed to define the transition and observation models for the dynamic Bayesian network.