A Novel Approach for Leak Localization in Water Distribution Networks Using Computational Intelligence

Abstract

This article provides a new approach to localize leaks in water distribution networks (WDN). The model-based leak localization is formulated as an inverse problem solved by using optimization tools. In most approaches to leak localization currently described in the scientific literature, the performance depends on the labeling method for nodes in the parameter space of objective functions. The main contribution of this article lies in a new model-based leak localization approach for water distribution networks whose performance does not depends on the node labeling method. As a result, the accuracy and efficiency of the optimization tools used for leak location task are improved. In addition, the proposal does not depend on the sensitivity matrix used in other methods to localize leaks. In order to achieve all the above, the method hereby proposed builds upon a localization strategy based on the network topological configuration. The Hanoi WDN is used to validate the proposed methodology where the metaheuristic algorithms differential evolution, particle swarm optimization, and simulated annealing are used as optimization tools.