Localization of Leaks in Water Distribution Networks

人们普遍认为，在北美的城市中，近三分之一的饮用水在处理和输送过程中丢失。这会导致宝贵的自然资源的损失，以及涉及储存、处理和运送的作业所浪费的能源。当世界各地的数百万人无法获得清洁饮用水时，这一点尤其令人担忧。人工智能(AI)和高级概率信号处理技术的发展显示出在解决水分损失方面的巨大潜力，特别是在及时检测和定位泄漏方面，否则这些泄漏可能会无限期地运行，而不会使用传统方法检测到。我们的工业合作伙伴已经投资了一项专利监测系统，该系统可以安装在消防栓上，以检测和定位泄漏，即使在寒冷的气候下也是如此。然而，决策支持工具仅限于泄漏检测，关键的泄漏定位方面尚未开发。该项目旨在引入人工智能和概率建模的最新发展，以准确定位(精确定位)复杂供水系统中的泄漏。通过这一合作关系，学术团队将利用实验室和现场数据开发和验证新的泄漏定位算法和流程，然后将其转化为现实世界复杂的操作环境。这项工作将带来先进的漏水定位能力，有助于公用事业公司及时检测和修复漏水的水管，从而将宝贵资源的损失降至最低。此外，学员还将获得在概率建模、人工智能和信号处理方面的实用技能，这些技能在加拿大需求很大。

It is widely believed that nearly a third of drinking water is lost between treatment and delivery in urban cities in North America. This results in the loss of a valuable natural resource and wasted energy from operations involving storage, treatment, and delivery. This is particularly concerning when millions around the world lack access to clean drinking water. Developments in artificial intelligence (AI) and advanced probabilistic signal processing techniques show significant potential in addressing water loss, specifically in timely detection and localization of leaks that otherwise could run indefinitely without being detected using conventional methods. Our industrial partner has invested in a patented monitoring system which can be installed on fire hydrants to detect and locate leaks, year-round even in cold climates. However, the decision-support tools are limited to leak detection, with the crucial leak localization aspect yet to be developed. This project aims to bring latest developments in AI and probabilistic modelling to accurately localize (pinpoint) leaks in complex water distribution systems. Through this partnership, new algorithms and processes for leak localization will be developed and validated by the academic team using laboratory and field data and then translated to real-world complex operational environments. This work will lead to advanced leak localization ability, which will assist utilities in timely detection and repair of leaky water mains, thus minimizing the loss of a valuable resource. As well, trainees will also gain practical skills in probabilistic modelling, AI, and signal processing which are in high demand in Canada.