Optimizing Water Distribution through Explainable AI and Rule-Based Control

Optimizing water distribution both from an energy-saving perspective and a quality of service is challenging task since it involves complex system with many nodes, hidden variables operational constraints. For this reason, systems need to handle delicate trade-off between the effectiveness computational time solution. In paper, we propose new computationally efficient method, named rule-based control, optimize networks without for rigorous formulation optimization problem. As matter fact, based on machine learning approach, proposed method employs only set historical data, where configuration can be labeled according criterion. Since data-driven could applied any network data are available. particular, control exploits classification that allows us retrieve rules leading good or bad performances system, even information about its physical laws. The evaluation results some simulated scenarios shows approach able reduce energy consumption while ensuring service. currently used in Milan (Italy) main.