Knowledge- and data-driven intelligent simulation method for tidal river network hydrodynamics

The intelligent and efficient joint scheduling of numerous sluices and pumping stations in tidal river networks is an important guarantee for improving water quality through increasing water mobility in these systems. However, previous intelligent simulation methods lack physical interpretability, making it difficult to accurately describe the complex hydrodynamic process in tidal river networks. This paper proposes a knowledge- and data-driven intelligent simulation method for tidal river network, which is applied to the hydrodynamic simulation of a generalized tidal river network and the Wennan part of the Shanghai tidal river network. This method takes artificial neural network (ANN) as the backbone and incorporates the river network flow control equations as physical constraints, and an ANN loss function is constructed by including the residuals of the control equations. The ANN weight set is iteratively optimized until the loss function meets the required criteria, thereby achieving a hydrodynamic intelligent simulation of tidal river network with both physical interpretability and high computational efficiency. This method differs from traditional ANN approaches in that it greatly reduces the amount of training data required and can also predict the hydrodynamic processes in sections without training data. This method has good simulation accuracy, computational efficiency, and robustness.