Pipe leakage detection system with artificial neural network

Abstract

This project aims to develop a system that can monitor to detect leaks in water distribution networks. It has been projected that leakage from pipelines may lead to significant economic losses and environmental damage. The loss of water from leaks in pipeline systems accounts for a large portion of the water supply. Pipelines are maintained throughout their lives span; however, it is difficult to avoid a leak occurring at some point. A tremendous amount of water could be saved globally if automated leakage detection systems were introduced. An embedded system that monitors water leaks can efficiently aid in water conservation. This project focuses on developing a real-time water leakage detection system using a few types of sensors: water flow rate sensor, vibration sensor, and water pressure sensor. The data from the sensors is uploaded and stored by the microcontroller (NodeMCU V3) to the database cloud (Google Sheets). The data that is stored in the database is analyzed by artificial neural network (ANN) by using Matlab software. An application is developed based on results from ANN training to detect the leakage event. Implementing the proposed system can increase operations efficiency, reduce delay times, and reduce maintenance costs after leaks are detected.