Subspace-Based Identification Techniques for a ‘Smart’ Wind Turbine Rotor Blade: A Study Towards Adaptive Data-Driven Control

More cost-effective wind turbines with higher rated power, are desired to emerge in the near future. The current trend in wind energy is for this reason the increase of the wind turbine dimensions. Increasing their size is however strongly limited by the rotor blades structural integrity towards occurring aeroelastic loads, especially with respect to blade fatigue. Advanced (re)active load reduction methods and advanced control techniques for wind turbines are therefore considered important research areas. The thesis provides a study towards subspace-based adaptive data-driven control for a ‘smart rotor blade. In an on-line setting closed-loop subspace identification can be of most importance to obtain adequate data-driven controllers for wind turbine installations, that are self-learning/adaptable to changing site/wind conditions. Moreover, the procedure of closed-loop subspace identification, followed by controller design in an off-line (automated) controller synthesis is also most desirable to find an adequate controller for a certain site location. The contents of the thesis are two-fold. It provides a simulation model of a 2-dimensional ‘smart rotor blade. This model effectively describes the unsteady aeroelastic behavior of a wind turbine rotor blade section for changing wind conditions, equipped with a trailing edge flap, and suitable to be interfaced with a self-learning adaptive controller. The thesis contributes furthermore by proposing several novel closed-loop Subspace Model Identification (SMI) techniques, which are presented in a unifying framework that also includes some recent state-of-the-art closed-loop SMI techniques from literature. Included is a comparative simulation study of these closed-loop SMI procedures based on several models, like the constructed ‘smart rotor blade model. In addition to obtaining insight in identification of such a controlled smart rotor blade for wind turbine purposes, the identified open-loop system model can also be used to predict the systems classical flutter instability point, especially in an on-line scheme. M.Sc. Thesis R.B.C. de Korte