A decentralized model predictive control scheme for the automatic generation control problem in power systems

Abstract

We present the application of decentralized Model Predictive Control (MPC) without terminal constraints to the Automatic Generation Control (AGC) problem in power systems. A dissipation-based constraint is additionally enforced to each MPC optimization of a control area — a subsystem within an interconnected system, to assure the overall system stability. A systematic tuning procedure for the enforced dissipation-based constraint is derived with a sequence of linear matrix inequality (LMI) optimizations. It is shown in numerical simulations that, not only the closed-loop stability, but the desirable performance of the AGC relatively to that of the centralized MPC is also obtained with the presenting decentralized MPC scheme.