ABSTRACT
Quasi-periodic loading resulting from waves and a rotationally sampled wind field often leads to fatigue-driven designs for offshore wind turbine support structures. The uncertainty on wind and wave loading, together with large modelling uncertainties, lead to large discrepancies between the observed and predicted dynamic behaviour of these structures. Among many recent-developed techniques for monitoring of true fatigue damage development, two promising Kalman-type filters are compared, namely the recently proposed Dual Kalman filter (DKF) and the Gillijns and De Moor filter (GDF). The filters are applied to synthetic vibration data in order to predict the global response of a lattice support structure assuming large modelling uncertainties and no knowledge of the input forces. A critical assessment of both filters with regard to requirements on the available data and tuning of the filter parameters is presented.
