ABSTRACT
Described herein is a method for the detection of pipe breaks and water losses in urban water distribution networks, by use of a wavelet change-point anomaly detection algorithm and streaming water consumption data from an urban locale. The wavelet change-point method utilizes the Continuous Wavelet Transform (CWT) of signals to analyze how the frequency content of a signal changes over time, and wavelet coherence to reveal time-varying frequency content common in multiple signals. The method also utilizes streaming water consumption data from consumers (‘automatic meter reading’ devices, AMR) and from District Meter Areas (DMA), to acquire inherent knowledge of water consumption at normal conditions at house and area-wide levels, and to make inferences about water consumption under abnormal conditions. This temporal anomaly detection is then georeferenced and used for spatial anomaly detection, producing ‘heatmap’ representations of the areas in the city with high probability of waterloss incidents.
