BROWSE
Detail
Conference Paper
Development of Decision Support Systems for Water Pipe Leakage Management: Anomaly Detection
Cited 0 time in 
Share 
Share
- Authors
- Hahyeon Moon, Sunwoo Kim, Hangyeol Seon, Kwang-Ju Kim, Mi-Sun Kang, Junghyun Kim
- Issue Date
- 2024-08
- Citation
- International Conference on Platform Technology and Service (PlatCon) 2024, pp.252-256
- Publisher
- IEEE
- Language
- English
- Type
- Conference Paper
- Abstract
- To enhance the water leakage detection system in South Korea, previous studies have introduced machine learning-based approaches. While these studies present well-developed models, it is important to note that there is still a gap especially related with computational efficiency and accuracy. This research proposes solutions to address computational cost associated challenges. First, this research presents the outlier detection method using autoencoder and CNN, highlighting their advantages over the previous work (i.e., Isolation Forest-based approach). Second, this research proposes a methodology to restructure line-based data into a tree format to reduce neighbor search times and computational costs. The results show that autoencoder and CNN achieve higher accuracy in outlier detection compared to the Isolation Forest algorithm. In addition, the tree-structured data significantly reduces computational costs. It seems that the proposed methods improve data quality and reduce computational costs, contributing to the operational efficiency of water pipe leakage management systems with potential applications in various public infrastructure systems.
- KSP Keywords
- Based Approach, Computational efficiency and accuracy, Data Quality, Detection Systems(IDS), Infrastructure System, Learning-based, Line-Based, Management system, Neighbor Search, Operational efficiency, Outlier Detection method