Genetic algorithm-based framework for optimal sensor placement in acoustic leak localization of water distribution networks

Accurately localizing leaks in water distribution networks (WDNs) remains a critical and ongoing research challenge. Acoustic-based methods are widely utilized for leak localization due to their higher sensitivity in detecting leaks and their reliance on sensor data rather than physical models. However, the problem of optimal sensor placement (OSP) for acoustic sensors has yet to be fully addressed, as robust analytical or numerical models for acoustic wave propagation in fluid-filled, two-dimensional (2-D) pipe networks are not readily available. Most existing OSP approaches rely on underlying hydraulic models, which are challenging to apply in acoustic sensing contexts. To address this gap, we propose a novel framework for solving the OSP problem specifically for acoustic sensors in WDNs. The framework defines an objective function based on “interior points”—specificlocations within the network where leaks can be accurately identified using a given sensor placement configuration. The optimization problem is solved using a Genetic Algorithm, incorporating constraints to account for sound attenuation, ensuring that leak signals remain detectable at the optimally placed sensors. This study provides valuable insights for water utilities, engineers, and other stakeholders, enabling them to make budget-conscious and informed decisions about the number and placement of acoustic sensors.