Online Detection of Water Contamination Under Concept Drift

TL;DR

This paper presents AD&DD, an unsupervised real-time water contamination detection method using dual-threshold drift detection and LSTM-VAE, effectively handling sensor drift and enabling decentralized monitoring in water networks.

Contribution

Introduces AD&DD, a novel unsupervised approach combining drift detection with LSTM-VAE for real-time water contamination detection under concept drift.

Findings

AD&DD outperforms existing methods in detecting anomalies.

Effective identification of sensor drift as concept drift.

Decentralized architecture enables accurate detection and localization.

Abstract

Water Distribution Networks (WDNs) are vital infrastructures, and contamination poses serious public health risks. Harmful substances can interact with disinfectants like chlorine, making chlorine monitoring essential for detecting contaminants. However, chlorine sensors often become unreliable and require frequent calibration. This study introduces the Dual-Threshold Anomaly and Drift Detection (AD&DD) method, an unsupervised approach combining a dual-threshold drift detection mechanism with an LSTM-based Variational Autoencoder(LSTM-VAE) for real-time contamination detection. Tested on two realistic WDNs, AD&DD effectively identifies anomalies with sensor offsets as concept drift, and outperforms other methods. A proposed decentralized architecture enables accurate contamination detection and localization by deploying AD&DD on selected nodes.