Differentiable neural network-based models enable gradient-based optimization for model predictive control of urban drainage networks.

Model predictive control (MPC) improves the use of the hydraulic capacity of an urban drainage network with online optimization, thus reducing overflow and flooding. However, its applicability is hindered by the heavy computational burden of repetiti…

Preview only. Read the full abstract at the source