Contamination source identification in water distribution network with deep utilization of monitoring information

Abstract:

To achieve timely and accurate identification of contamination source in water distribution networks, a sequential Bayesian method was proposed specifically for water distribution systems equipped with online water quality monitoring devices. The method utilized temporal information from water quality sensors to identify the contamination source in water distribution network with stochastic fluctuation in water demand. Monte Carlo simulations were conducted to generate contamination events and establish the observation probability distribution function for each node. Then this information was used to compute the posterior probability of each possible source for the observed alarm pattern in real time by using Bayesian inference. Finally, the contamination source was identified by ranking the posterior probabilities. Furthermore, the influence of different utilizations of sensor information on the identification results was also compared. The results show that the proposed method enables continuous updating of the posterior probabilities of suspicious nodes when sensor information is gathered, resulting in fewer candidate nodes and lower information entropy. The method can effectively identify the contamination source, and the accuracy of contamination source identification improves with deeper utilization of sensor information. Introduction of initial alarm time as auxiliary information can reduce the number of candidate nodes and reduce the uncertainty in probability distribution of suspicious contamination nodes, thus improving the accuracy of identification.