A Bayesian Approach for Probabilistic Contamination Source Identification

摘要

Drinking water distribution system models have been prominent in the development and implementation of contaminant warning systems. This study proposes a new probabilistic contaminant source identification algorithm using a Beta-Binomial conjugate pair framework to identify contaminant sources in water distribution system, and compares the performance of this algorithm to a previous study using a discrete probability representation based on Bayes' Rule. The evaluation of the performance associated with the two algorithms was conducted using a simulation study with a conservative "chemical injection" event within a small distribution system network. Preliminary results showed that while the Bayes' Rule approach responded faster, the algorithm can quickly become insensitive to changes in the event detection signal. However, the Beta-Binomial approach appeared to better represent the true source location and injection time.