Subspace predictive repetitive control to mitigate periodic loads on large scale wind turbines

Manufacturing and maintenance costs arising out of wind turbine dynamic loading are one of the largest bottlenecks in the roll-out of wind energy. Individual Pitch Control (IPC) is being researched for cost reduction through load alleviation; it poses a challenging mechatronic problem due to its multi-input, multi-output (MIMO) nature and actuation constraints related to the wear of pitch bearings. To address these issues, Subspace Predictive Repetitive Control (SPRC), a novel repetitive control strategy based on the subspace identification paradigm, is presented. First, the Markov parameters of the system are identified online in a recursive manner. These parameters are used to build up the lifted matrices needed to predict the output over the next period. From these matrices an adaptive repetitive control law is derived. To account for actuator limitations, the known shape of wind-induced disturbances is exploited to perform repetitive control in a reduced-dimension basis function subspace. The SPRC methodology is implemented on a high-fidelity numerical aeroelastic environment for wind turbines. Load reductions are achieved similar to those obtained with classical IPC approaches, while considerably limiting the frequency content of the actuator signals.