Combining machine learning and operational modal analysis approaches to gain insights in wind turbine drivetrain dynamics

Abstract

Operational Modal Analysis (OMA) is one of the branches of experimental modal analysis which allows extracting modal parameters based on measuring only the responses of a structure under ambient or operational excitation which is not needed to be measured. This makes OMA extremely attractive to modal analysis of big structures such as wind turbines where providing measured excitation force is an extremely difficult task. One of the main OMA assumption concerning the excitation is that it is distributed randomly both temporally and spatially. Obviously, closer the real excitation is to the assumed one, better modal parameter estimation one can expect. Traditionally, wind excitation is considered as a perfect excitation obeying the OMA assumptions. However, the present study shows that the aeroelastic phenomena due to rotor rotation dramatically changes the character of aerodynamic excitation and sets limitations on the applicability of OMA to operational wind turbines. The main purpose of the study is to warn the experimentalists about these limitations and discuss possible ways of dealing with them.