Aggregation-based model predictive control of open channel networks

By use of the linearized Saint-Venant equations, this paper first recalls the so-called constant volume model of open channel networks. Based on this model, a predictive model of open channel networks is derived. It takes the storage volume variations as state variables and gate openings as control variables with constraints. Considering the structure complexity of open channels and constraints, the control problem of open channel networks is cast in the framework of model predictive control. An equivalent aggregation scheme for the model without any constraints is presented to improve the computational efficiency significantly without any loss of control performance. Then the idea of equivalent aggregation scheme is generalized to the case with constraints and a quasi-equivalent aggregation scheme is developed. A detailed study for a particular open channel network consisting of two pools is illustrated to demonstrate the efficacy of the aggregation-based model predictive control algorithm.