Integrated data-driven framework for anomaly detection and early warning in water distribution system

摘要

Data-driven anomaly detection and early warning have been extensively used in water distribution systems (WDS). Events such as pipe bursts and sensor failure cause abnormal monitoring data. Anomaly detection during real-time data monitoring and identification of various events are crucial in WDS. This study proposes a framework for anomaly detection and early warning in WDS. This framework comprises four anomaly detection modules-single-point anomaly identification, sensor sequence, inter-sensor sequence, qualitative module. A case study is conducted using the Net3 pipe network model. The results indicate that the proposed method can accurately identify pipe bursts and detect situations causing abnormal sensor data.

关键词