Optimal control of wind power systems with energy storage

We investigate how a wind power plant with energy storage can be optimally controlled to comply with power gradient restrictions. For this purpose we use the kinetic energy stored in the turbine rotor and an additional battery storage. As a main contribution we propose a two level control scheme combining a supervisory controller with a low level controller. A model predictive controller acts as a supervisory controller and provides set-points for a low level proportional integral controller. The low level controller applies its input to a higher order wind turbine model in real-time. As a result, the proposed control scheme substantially mitigates power gradients, while keeping energy losses and increased turbine loading to a minimum. For a battery power of 20% the rated wind turbine power, we showed that the maximum power gradient is reduced by 78.5% and the standard deviation by 88.7% compared with a conventional, gain-scheduled PI controller.