View article

This paper presents a review of methods to remove harmonic components from vibration signals acquired on machines with rotating elements. In particular, the issues associated with wind turbine analysis will be presented and investigated. Although Operational Modal Analysis is widely applied, the assumption of having white noise excitation cannot always be met. In particular, the presence of harmonics has proven to be a critical issue to be tackled. Different methodologies to separate harmonic and random components have been proposed in literature and will be analyzed in this paper. The performance of a selection of these algorithms will be analyzed firstly on simulated data for a particularly critical case. Afterwards, the data generated from a full-scale aeroelastic model of a wind turbine will be analyzed. The reference frequencies will be computed from the numerical model and compared with those obtained from Operational Modal Analysis in parked conditions. These will be used as reference to prove the efficiency of harmonic removal methods for operational data.