Efficient tuning of Individual Pitch Control: A Bayesian Optimization Machine Learning approach

With the trend of increasing wind turbine rotor diameters, the mitigation of blade fatigue loadings is of special interest to extend the turbine lifetime. Fatigue load reductions can be partly accomplished using individual pitch control (IPC), and is commonly facilitated by the so-called multiblade coordinate (MBC) transformation. This operation transforms and decouples the blade load signals in a non-rotating yaw-axis and tilt-axis. However, in practical scenarios, the resulting transformed system still shows coupling between the axes. To cope with this phenomenon, earlier research has shown that the introduction of an additional MBC tuning variable - the azimuth offset - decouples the multivariable system. However, the introduction of this extra variable complicates the controller design process, and requires expert knowledge and specialized analysis software. To provide an efficient method for the optimization of fixed-structure IPC controllers, based on black box and computationally costly objective functions, this paper considers a Bayesian optimization controller tuning framework. Results show the efficiency of the framework to tune a combined 1P+2P IPC implementation, without prior knowledge, and based on high-fidelity simulation results using a computationally expensive objective function.