Wipro Limited
Method and system for predicting water leakage events in district-metered-area of water distribution network

Abstract:

Disclosed herein is a method and system for predicting water leakage events in district metered area of water distribution network. The method comprises receiving historic data of water flow and water pressure associated with DMA for predicting flow data and pressure data of water in the DMA for predefined future time period. Thereafter, method comprises retrieving plurality of cluster points corresponding to the predicted flow data, the pressure data, and an anomaly score identified for each of the plurality of cluster points from machine learning model. The method detects one or more cluster points as anomaly points by comparing anomaly scores with lower bound value and upper bound value and thereafter predicts leakage events corresponding to each of the detected anomaly points. In this manner, present disclosure predicts water leakage events in DMA for future time intervals and hence necessary leakage maintenance actions may be taken to avoid such events.