LAUSR

To harvest wind energy from nature, wind turbines are increasingly installed globally, and the blades are the most essential components within the turbine system. The blades usually suffer from time-varying nonstationary wind loads, and the load information is normally unknown or difficult to collect. This poses significant challenges to the blade assessment and damage detection. Transmissibility function (TF) methods have the potential to address this challenge as they do not require loading information. In this article, a novel wavelet package energy TF (WPETF) method is proposed to increase the high-frequency resolution while maintaining its low sensitivity to noise, and it is further used for wind turbine blade fault detection. Compared with the existing Fourier TF method, the proposed method is immune to the external loading impacts, does not require excitation knowledge, and is robust to noise. Compared with the existing wavelet energy TF method, the novel one uses wavelet package decomposition instead of wavelet decomposition to further increase the high-frequency resolution, which provides richer damage-induced information. The effectiveness of the WPETF method for wind turbine blade condition assessment is first verified numerically, and then on three industrial-scale wind turbine blades with both naturally (uncontrolled) and artificially introduced (controlled) damage. Its advantages over a number of existing methods are also demonstrated.