%0 Journal Article %A Wu, Yipeng %A Liu, Shuming %D 2017 %T CCWI2017: F7 'Clustering-based Burst Detection Using Multiple Pressure Sensors in District Metering Areas' %U https://orda.shef.ac.uk/articles/journal_contribution/CCWI2017_F7_Clustering-based_Burst_Detection_Using_Multiple_Pressure_Sensors_in_District_Metering_Areas_/5363869 %R 10.15131/shef.data.5363869.v1 %2 https://orda.shef.ac.uk/ndownloader/files/9218359 %K Burst Detection %K Cosine Distance %K Clustering %K CCWI2017 %K Civil Engineering not elsewhere classified %X Bursts in water distribution systems (WDS) can cause water loss, service interruptions and other negative effects. However, it is challengeable for worldwide water utilities to timely be aware of bursts. This paper presents a novel burst detection approach using data from multiple pressure sensors in district metering areas (DMA). Differing from most data-driven methods that employ prediction models, this method utilizes a clustering algorithm to detect burst-induced data. Owing to the use of cosine distance in clustering analysis, temporal varying correlation between data from diffierent sensors is exploited, making the method only requires one day’s worth of data to implement. When applied to a DMA with three pressure sensors, the method successfully detected some real and simulated bursts over a period of two months.
%I The University of Sheffield