A flexible, leak crew focused localization model using a maximum coverage search area algorithm

• Published in: IOP conference series. Earth and environmental science Vol. 1136; no. 1; pp. 12042 - 12051
• Main Authors: Snider, Brett, Lewis, Gareth, Chen, Albert, Vamvakeridou, Lydia, Savić, Dragan
• Format: Journal Article
• Language: English
• Published: Bristol IOP Publishing 01.01.2023
• Subjects: Algorithms; Clustering; Crews; Localization; Machine learning; Pipes; Searching; Utilities
• ISSN: 1755-1307; 1755-1315; 1755-1315
• DOI: 10.1088/1755-1315/1136/1/012042

Buried watermains are deteriorating and pipe failure is increasing in many cities. In response, advanced leak location models have been developed to help identify where a leak is occurring – which allows utilities to react quickly to pipe bursts and reduce the impact of the leak. This paper develops a new leak location model that is designed to identify optimal search areas for leak crews using a random forest classification model and the maximum coverage location problem algorithm. The model, when compared with other machine learning and clustering localization predictions, reduces the search space by over 35%, allowing utilities to confirm leak location and mitigate its impact more efficiently. The new model is also highly customizable, able to adjust the number of search areas and search size quickly and easily to meet leak crews’ requirements.