Uncertainty propagation and subspace identification of isotropic wind turbines with the PO-MOESP algorithm

The estimation of modal parameters of an operating wind turbine is an important task, that can be carried out using subspace algorithms. The PO-MOESP is one of the subspace algorithms suitable for this task, since it can model the atmospheric turbulence as a process noise. Besides the point estimates available through the subspace identification, the knowledge of the related uncertainty bounds is fundamental for structural monitoring. While the methodology for uncertainty quantification is well-established for many subspace algorithms, the dedicated uncertainty propagation for MOESP-type of methods is missing. The aim of this work is to close this gap and develop an uncertainty quantification scheme suited for PO-MOESP. The method is validated by Monte Carlo simulations of the IEA 10 MW reference turbine in turbulent wind, proving its accuracy with a difference of less than 10 % between the target and the estimated modal parameter covariances. The application to a dataset recorded on the Aventa AV-7 turbine confirms the accuracy of the developed approach.