ZhangJuan,ZhuJianxuan,LiuShuming,et al.Anomaly Detection of Leakage Rate in DMA Districts Based on Heuristic Gaussian Cloud Transform[J].China Water & Wastewater,2026,42(17):48-54.
Anomaly Detection of Leakage Rate in DMA Districts Based on Heuristic Gaussian Cloud Transform
China Water & Wastewater[ISSN:1000-4062/CN:12-1073/TU]
volume:
第42卷
Number:
第17期
Page:
48-54
Column:
Date of publication:
2026-09-01
- Abstract:
- To address the shortcomings of traditional anomaly detection methods for leakage rates in district metered areas (DMA), which rely on fixed thresholds or empirical values, this paper proposes a dynamic threshold determination method based on heuristic Gaussian cloud transform (H-GCT). The method employs a Gaussian mixture model (GMM) to fit historical leakage rate data, and further combines it with a Gaussian cloud model to quantify the concept ambiguity and the degree of overlap, thereby enabling the classification of normal and abnormal leakage rates and achieving more scientific anomaly detection. Data from 30 DMA districts in city L were selected for validation. The results showed that the threshold leakage rates ranged from 3.5% to 9.0%, with an average concept ambiguity of 0.341 4, which was consistent with the actual operational characteristics of the water distribution network. The proposed method can provide a scientific basis for leakage monitoring and zonal metering in water supply pipeline networks.
Last Update:
2026-09-01