A comparative study of time-frequency representations for fault detection in wind turbine

To reduce the cost of wind energy, minimization and prediction of maintenance operations in wind turbine is of key importance. In variable speed turbine generator, advanced signal processing tools are required to detect and diagnose the generator faults from the stator current. To detect a fault in non-stationary conditions, previous studies have investigated the use of time-frequency techniques such as the Spectrogram, the Wavelet transform, the Wigner-Ville representation and the Hilbert-Huang transform. In this paper, these techniques are presented and compared for broken-rotor bar detection in squirrel-cage generators. The comparison is based on several criteria such as the computational complexity, the readability of the representation and the easiness of interpretation.