Abstract
Model-based and data-driven methods are commonly used in leak location strategies in water distribution networks. This paper formulates a hybrid methodology in two stages that complements the advantages and disadvantages of data-driven and model-based strategies. In the first stage, a support vector machine multiclass classifier is used to reduce the search space for the leak location task. In the second stage, leak location task is formulated as an inverse problem, and solved using a variation of the differential evolution algorithm called topological differential evolution. The robustness of the method is tested considering measurement and varying demand uncertainty conditions ranging from 5 to 15% of node nominal demands. The performance of the hybrid method is compared to the support vector machine classifier and topological differential evolution approaches as standalone methods of leak location. The hybrid proposal shows higher performance in terms of location accuracy, zone size, and computational load.
Highlights
Water, a vital resource for humanity, plays multiple roles in everyday life: ranging from drinking water consumption to the fulfilment of daily tasks
The main goal of this work, and its main contribution, is the proposition of a hybrid methodology for the location of leaks in water distribution networks (WDNs). This methodology consists of two stages: an initial data-driven stage in which a subzone of the WDN is identified as a potential leak location by means of a multiclass support vector machines (SVMs) classifier; and a second stage that improves the leak location estimated in the first stage through a model-based approach
The estimated leak location is generated in two consecutive stages: On the first stage, the WDN is partitioned into a set of candidate leak zones using agglomerative clustering [29], and taking into account the topological relationships between network nodes, presented in Λ ∈ < NT × NT
Summary
A vital resource for humanity, plays multiple roles in everyday life: ranging from drinking water consumption to the fulfilment of daily tasks. Chen et al [13] and Zhang et al [20] use clustering algorithms based on sensitivity matrixes generated by network models to group the network nodes into zones before locating the leaks. Quinones-Grueiro et al [11], use k-medoids clustering to group the nodes according to their topological characteristics (shortest pipe distances between network nodes) Another zone generation approach consists in selecting multiple ranking candidates from a final solution. This methodology consists of two stages: an initial data-driven stage in which a subzone of the WDN is identified as a potential leak location by means of a multiclass SVM classifier; and a second stage that improves the leak location estimated in the first stage through a model-based approach This second stage is formulated as an inverse problem, and it is solved using a variation of the Differential.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
