Cloud-Based Water Leakage Detection and Localization

In this paper, a cloud-based system for smart water management is proposed to detect the water leakage and to predict the location of leakage in pipes. The system utilizes the flow rates of water in pipelines to determine leakages and applies machine learning (ML) techniques to predict the location of the leakages. A hybrid ML model has been developed, combining a multi-layer perceptron (MLP) and a support vector machine (SVM), which is used to predict the location of the leakages in the pipeline. To minimize the losses of water, an itinerary for leak maintenance is proposed, which prioritizes leaks based on its location and the extent of the outflow. To test the proposed system, a prototype has been developed and the same is modelled in STAR-CCM+, a Computational Fluid Dynamics (CFD) software. The results show that the leakage detection algorithm has an accuracy of 99% while location prediction using machine learning has an accuracy of 94.14%.