A methodology for leak detection in water distribution networks using graph theory and artificial neural network

Considering the scarcity of water resources, it is necessary to identify the leakage in Water Distribution Networks (WDNs). In this paper, a step-by-step method of WDN decomposition has been introduced for leak detection. First, the WDN is divided into two parts using the graph theory, then the part with leakage is identified using the results of pressure loggers and the artificial neural network. This process continues for the identified part to reach the limited leakage area. This method was applied to the Balerma WDN with five leakage scenarios including uncertainty of demand and pressure parameters. The results show that the proposed method can find the leakage area of WDNs with good accuracy.