Bayesian networks for estimating contaminant source and propagation in a water distribution system using cluster structure

Bayesian belief networks are a probabilistic analysis tool for representing and analyzing problems involving uncertainty. The problem of monitoring the propagation of a contaminant in a water distribution system can be naturally represented using Bayesian networks (BN). The presented methodology proposes estimating the likelihoods of the injection location of a contaminant and its propagation in the system using BN statistics. A clustering method, previously developed by the authors, is first applied to formulate a simplified representation of the distribution system resulting in an aggregated system. The aggregated network is represented as a directed acyclic graph and facilitates in the construction of a legal BN. The data collected from monitoring stations located at any of the nodes of the system is exploited to inquire about the possible sources of contamination and the consequent polluted nodes. For small networks, the probabilities can be estimated using exact inference algorithms, for large networks - using approximated inference algorithms such as likelihood weighting. The proposed methodology is developed and tested on two water supply systems. The results demonstrate a promising potential of the proposed method.