JIANCai,GAOJin-liang,XUYong-peng.Iterative Partition Identification Algorithm for Anomaly Isolation of Water Distribution Network[J].China Water & Wastewater,2022,38(23):31-37.
Iterative Partition Identification Algorithm for Anomaly Isolation of Water Distribution Network
China Water & Wastewater[ISSN:1000-4062/CN:12-1073/TU]
volume:
第38卷
Number:
第23期
Page:
31-37
Column:
Date of publication:
2022-12-01
- Keywords:
- water distribution network; anomaly isolation; iterative partition; anomaly identification; convolutional neural network
- Abstract:
- To minimize abnormal event risks in urban water distribution network (such as pipe burst, leakage and illegal water use), a method for timely and effective isolation of abnormal events was proposed, which included two steps: partition and identification. First, the pipe network was divided into two sub-regions by combination graph theory technology, clustering method and manual fine tuning. Then, the specific occurrence area of abnormal events was determined by convolutional neural network algorithm and hydraulic simulation. The partitioning and identification process iterated until preset stop conditions were met, so as to isolate the abnormal event in the smallest possible area. To improve the computational efficiency, the subgroup technique was used in the hydraulic simulation process, and the most important monitoring data vector features were evaluated and selected in the training convolutional neural network algorithm. The results of case analysis showed that the proposed method was capable of isolating network anomalies reliably and efficiently, and had a good performance in terms of accuracy and timeliness.
Last Update:
2022-12-01