An extension of the modified constitutive relation error concept for the robust multi-physics offshore wind turbine model calibration from in situ data

In order to address health monitoring issues on offshore wind turbines, while taking into account engineering know-how and constraints, we propose a robust model identification procedure based on reliability of information for such structures. The procedure leans on the modified Constitutive Relation Error concept, which naturally considers sensing and modeling uncertainties. The concept is here extended to a multi-physics framework involving nonlinear fluid–structure interactions. In this framework, each individual interface is seen as a specific and potentially nonlinear constitutive relation, assumed unreliable and relaxed in the identification process, with suited error metric. Parameters associated to these constitutive relations are then identified to close at best the gap between operational observations and model predictions, and to detect modeling incompatibilities by means of error indicators. Considering classical engineering modeling and simulation approaches dedicated to offshore wind turbines, with inherent idealization and model bias, the performance of the approach is numerically analyzed. This is done first on academic examples with synthetic data, then in a more practical context by processing real in situ measurements coming from an industrial demonstrator equipped with various multiphysic sensors. Results notably illustrate the capability of the approach to cope with crude models by means of its bias-aware formulation.