Contamination Source Identification in Water Distribution Systems Using an Adaptive Dynamic Optimization Procedure

Abstract

Contamination source identification involves the characterization of the contaminant source based on observations that stream from a set of sensors in a water distribution system (WDS). The streaming data can be processed adaptively to provide an estimate of the source characteristics at any time once the contamination event is detected. In this paper, an adaptive dynamic optimization technique (ADOPT) is proposed for providing a real-time response to a contamination event. A new multiple population–based search that uses an evolutionary algorithm (EA) is investigated. To address nonuniqueness in the initial stages of the search and prevent premature convergence of the EA to an incorrect solution, the multiple populations are designed to maintain a set of alternative solutions that represent various nonunique solutions. As more observations are added, the EA solutions not only migrate to better solution states but the number of solutions decreases as the degree of nonuniqueness diminishes. This new algorithm adaptively converges to the solutions that best match the available observations. The use of the developed method is demonstrated for two WDS networks.