Volume 26: Closing Carbon Cycles – A Transformation Process Involving Technology, Economy, and Society: Part I

Pipeline Block Localization in Water Distribution Networks Using Artificial Neural Network Dinghuang Xing, Guang Yang, Hai Wang

Abstract

The pipeline blockage increases the resistance of water distribution networks. In this study, a new method based on machine learning was proposed to locate the blocked pipeline. Numerous block scenarios were simulated by the hydraulic simulation, considering various block sizes and user demands for each pipeline. The dataset of the pressure change rates on the nodes was used to train artificial neural network models. The influences of the dataset variables on the model performance were analyzed. Results showed that the proposed method can successfully locate the blocked pipeline using the measurement of one day.

Keywords water distribution networks, artificial neural network, block localization