Development of Smart Underground Drainage Leak Detection and Localization Using Acoustic and Flow Sensing with TINYML and IOT

Underground pipeline leakage leads to significant water loss, infrastructure damage, and economic impact, necessitating efficient detection and localization methods. This paper presents a smart leak detection system that integrates acoustic sensing, flow monitoring, Tiny Machine Learning (TinyML), and Internet of Things (IoT) technologies for real-time analysis. A hybrid sensing approach is employed, where flow sensors detect anomalies and trigger acoustic analysis using digital microphones. A lightweight TinyML model is deployed on an ESP32 microcontroller to classify leak signatures from background noise, reducing false alarms. Leak localization is achieved using the Time Difference of Arrival (TDOA) algorithm based on synchronized acoustic signals. Simulation studies using MATLAB Simulink and ANSYS validate the localization accuracy and fluid dynamics behavior. Experimental results demonstrate reliable leak detection with real-time cloud monitoring via Firebase. The proposed system offers a low-cost, energy-efficient, and scalable solution for smart water infrastructure monitoring.