A Novel Rule-Based Modeling and Control Approach for the Optimization of Complex Water Distribution Networks

Enrico Ferrari, Damiano Verda, Nicolò Pinna, Marco Muselli

Abstract: This work applies Rule Based Control, a new rule-based, computationally efficient machine learning method for optimizing complex networks. This approach does not require a rigorous formulation of the optimization model, since only a set of historical data, where solutions are labeled as good or bad, is needed. It makes use of a rule-based machine learning method, which allows the optimization of complex networks, where a full description of the system is not available or too complex. The proposed approach is currently under evaluation in the water distribution system of the Milan (Italy) water main. The application of the approach to synthetic data shows its ability of reducing the energy consumption, while ensuring a good quality of service.