Closed loop system identification of an industrial wind turbine system: experiment design and first validation results

Reliable dynamical models of complete wind turbine systems are expected to be of great importance for the development of economically attractive wind turbine systems. Reliability is closely related to validity, which is difficult to assess from measurements and model simulation only. This is mainly due to fluctuations of the wind field corrupting the the measurements with a noisy contribution. Application of system identification solves this problem because the noise model accounts for the (wind induced) disturbances and all attention can be focussed on the deterministic part describing the dynamics which are the subject of validation. An industrial full size wind turbine system is available for identification experiments. A portable PC is equipped with data acquisition hardware and is customized to allow for easy and flexible experimentation. Experiments were restricted to closed loop situation with existing single loop PI controllers. Plant models are identified from this closed loop data using a two step approach. In the validation stage the identified model compared reasonably well to the physical model of the wind turbine