Leak localization of water supply network based on temporal convolutional network
jie zhang, Xiaoping Yang, Juan Li
Abstract
Abstract In recent years, the urban water supply network system has faced severe challenges. The aging, corrosion, and manmade damage to pipelines waste a lot of water resources and cause harm to human beings. Therefore, this paper proposes a method for locating leak locations in a water supply network using temporal convolutional networks. First, a continuous sequence of pressure signals is input into the proposed network model. Then, we map it to two parallel outputs by the network model. In the first output, leak detection is performed as a multi-label classification task. In the second output, the location of the leak is determined using a regression algorithm. This paper tests the proposed network framework on benchmark networks. The results show that the network framework can obtain accurate leak locations and outperform the commonly used network frameworks.