Pipeline-Burst Detection on Imbalanced Data for Water Supply Networks

Data-driven methods based on samples from a supervisory control and data acquisition system have been widely applied in water-supply-network burst detection to save unexpected economic labor costs. However, the class imbalance problem actual on-site monitoring needs be revised improve performance of data-driven methods. In this study, we proposed domain adaptation method generate minor-category (pipeline-burst general) arbitrary pipe networks utilizing theoretical hydraulic models. The transferred pipeline-burst generated random water supply network with models an imbalanced dataset. Accordingly, established machine learning model exploring mapping matrix between two domains for minority-category transfer. experimental validation first verified effectiveness reliability customized terms their bust recognition accuracy, parameter sensitivity time efficiency. Then, dataset working was used prove suitability compatibility method. A bust-point location also provided results pipeline-bursting events. validations show superiority our approach detection.