Secondary Model Predictive Control Architecture for VSC-HVDC Networks Interfacing Wind Power

Abstract

This paper proposes a secondary Model Predictive Control (MPC) based architecture to provide DC voltage and DC power control to MT VSC-HVDC networks interfacing wind power generation. The proposed architecture places the controller at a supervisory level, thus providing coordination amongst existing converters' local droop controllers. Droop control is a type of decentralized DC voltage/power control. Hence, in case of secondary controller failure, droop control acts as a contingency control scheme. A simplified linear dynamic model of the MT VSC-HVDC network is utilised for the MPC's design, thus minimizing computational effort needed to compute secondary control action. Updates of droop gains and offshore and onshore “AC” variables, in the dq domain, are explicitly considered as measured input disturbances within the formulation of the MPC controller, thus, ensuring appropriate control response in order to minimise adverse impact of their variation on the overall system's performance, particularly under wind power variations. Simulation results show that MPC, whose design is based on the system's simplified linear model, is capable of delivering satisfactory performance when applied to the high fidelity non-linear full order model of a six-terminal VSC-HVDC network simulated in PSCAD.