Water distribution systems event detection through classification and regression trees

Event detection is one of the current most challenging topics in water distribution systems analysis: how regular on-line hydraulic (e.g., pressure, flow) and water quality (e.g., pH, free chlorine) measurements at different network locations can be efficiently utilized to detect accidental or deliberate water quality contamination events. This study deals with the estimation and classification of measured water quality data aimed at identifying possible contamination events. Regression and classification trees are utilized to estimate parameters' future data and classify outputs. Estimation is applied on routine water quality data and classification on simulated water contamination events. Estimation and classification were carried out for four water quality parameters: Cl, Temp, pH, and EC. Residuals were analysed using confusion matrices and ROC curves. Preliminary results show promising potential for efficient identification of water anomalies using the proposed methodology.