Application of Distributed and Hierarchical Model Predictive Control in Hydro Power Valleys

Abstract

We provide an overview on the application of distributed and hierarchical model predictive control (MPC) algorithms for the power reference tracking problem of the HD-MPC Hydro Power Valley (HPV) system (Savorgnan and Diehl, 2011). Serving as a case study for distributed and hierarchical MPC, the HPV benchmark has various challenging features, including nonlinear, non-smooth, and coupled cost function and nonlinear coupled subsystem dynamics. We propose different approaches to address these challenges and summarize our recently developed hierarchical and distributed MPC frameworks that could be applied to the HPV control problem. A comparison of distributed MPC based on a state-of-the-art distributed optimization method (Giselsson et al., 2012) with centralized and decentralized MPC is provided via numerical simulations. It is shown that by using a dynamic division of total power reference to deal with the coupling in the cost function and a specific formulation of the dual optimization problem, distributed MPC achieves almost the same tracking performance as centralized MPC, with the advantage of being implementable in a distributed setting.