LeakDB : A benchmark dataset for leakage diagnosis in water distribution networks
Place of publicationKingston, Ontario, Canada
SourceProceedings of the 2018 Joint Conference on Water Distribution Systems Analysis and Computing and Control for the Water Industry (WDSA/CCWI 2018)
1st International WDSA / CCWI 2018 Joint Conference
Google Scholar check
MetadataShow full item record
The increase of streaming data from water utilities is enabling the development of real-time anomaly and fault detection algorithms that can detect events, such as pipe bursts and leakages. Currently, there is not a widely accessible dataset of real or realistic leakage scenarios, which could be used as a common benchmark to compare different algorithms, as well as to support research reproducibility. In this work we propose the design of a realistic leakage dataset, the Leakage Diagnosis Benchmark (LeakDB). The dataset is comprised of a large number of artificially created but realistic leakage scenarios, on different water distribution networks, under varying conditions. Additionally, a scoring algorithm was developed in MATLAB to evaluate the results of different algorithms using various metrics. The usage of the LeakDB dataset, is demonstrated by scoring four detection algorithms. The dataset is stored on an open research data repository, and will be updated in the future with new simulation scenarios. The source code of the toolkit that generates the leakage benchmark dataset, as well as the detection algorithms used, are released as open source.