An inversion based model-predictive feed-forward discharge controller for impounded rivers

The automatic management of barrage controlled river segments tries to combine several contradicting goals. To facilitate ship traffic and at the same time prevent large water discharge surges, the two competing goals of a steady water level and a homogenized water discharge into the next river section have to be harmonized with each other. As local water level control is already state of the art, an approach to homogenize the resulting water discharge by utilizing water level tolerances is proposed in this work. To achieve this goal, a model predictive water level coordinator is proposed that uses an inverted distributed parameter river section model. The use of this coordinator on standard industrial control hardware with limited computational capabilities requires the simplification of the existing nonlinear Saint-Venant river models. A reduced linear Hayami model is therefor deducted and identified, making a power series based analytical solution possible which reduces the computational cost significantly. The effectiveness of this method will be illustrated and evaluated on a verified and precise model of an impounded river section of the lower Mosel river with a fully automated barrage system. It is shown that the water discharge can be homogenized while keeping the the water level within given bounds.