Operação inteligente e integração em tempo-real para a deteção de perdas de água

The objectives of the I-ReTiS-LeaksD&Op project is to innovate water supply systems (WSS) management through advanced computational tools and methodologies. This interdisciplinary effort focuses on the development of a Smart Predictive Digital Twin (SPDT) that encompasses a multifaceted approach to revolutionize WSS operational management through intelligent operations and leak detection and location and relies in two pioneering components: 1. Smart Predictive Digital Twin Development: 1. 1. The SPDT integrates a novel physics-informed machine learning (ML) model based on neural differential equations for WSS hydraulic modeling. This data-driven approach surpasses traditional models by eliminating scalability issues and the need for extensive calibration, offering quick prediction capabilities without being bound to specific mathematical formulations. 1.2. Includes an optimization module for energy and cost efficiency, focusing on variable-speed-pump operations, pressure control, dynamic energy tariff adjustments, and leveraging renewable energy sources. Although the demand-response paradigm is highly complex, its potential impact is equally significant. 1.3. The framework enables real-time communications in a multiservice context, with continuous feedback and error control to manage uncertainties inherent in predictive analytics. This is a significant leap forward for the water sector, technologically and scientifically, offering a new paradigm for real-time WSS management with enhanced robustness, operational planning, leak reduction and energy cost savings. 2. Advanced Leak Detection and Localization: 2.1. This component aims to refine real-time leak detection and localization within WSS, employing a dual approach that combines an auto-calibrated hydraulic model with a physics-informed ML digital twin. This innovative framework utilizes an active indirect model-based method, comparing real-time operational data with digital twin predictions to pinpoint leaks. 2.2. The synergy of a sensitive-based automatically calibrated hydraulic model and a physics-informed ML model offers a unique advantage. This approach reduces the reliance on large data sets and increases model accuracy and training efficiency by incorporating physical laws into the ML training phase. This method represents a breakthrough application of physics-based data analysis to WSS management. Both components will be tested in a real operational environment within a Portuguese WSS (~TRL 7), signifying a substantial step towards practical application. The outcomes of this project promise substantial advancements in WSS operational management by (i) enhancing energy costs efficiencies, (ii) improving real-time monitoring and control capabilities, (iii) expanding operational planning, and (iv) reducing water leaks and their associated environmental, financial, and social impacts. The scientific innovation of this project will be communicated through high impact international journals and includes (a) the Physics-Informed Machine Learning (ML) for WSS Hydraulic Modeling; (b) development of a Smart Predictive Digital Twin (SPDT); (c) Integration of Energy and Cost Optimization; (d) Real-Time Leak Detection and Localization using a Hybrid Methodology; (e) Application of Automatic Calibration and Sensitivity-Based Models and (f) Real-Time Communications in a Multiservice-Based Framework. Action on the target sector and development of promotional materials are also goals of th