Wind Turbine Blade Fault Detection Using Wavelet Power Spectrum and Experimental Modal Analysis

Abstract

The dynamic behavior of modern multi-Megawatt wind turbines has become an important design consideration. The power generated by wind turbine depends mainly on the interaction between the blades and the flowing wind. Therefore, the major aspects related to the operational reliability of a wind turbine are the aerodynamics characteristics and integrity of the blades. This paper demonstrates the results of the application of wavelet transform combined with experimental modal analysis as a tool for wind turbine blade condition monitoring. An experimental model analysis with different vibrational excitations performed for healthy and cracked cantilever stepped beam, which approximate the wind turbine blade when attached to the wind turbine hub. The beam response signals acquired for three different locations of vibration sensor. The cantilever beam with breathing crack response exhibits a non-stationary vibrational signal because of crack open-close cycles during the oscillation of the beam. The wavelet transform as a time-frequency signal processing technique is an effective tool to extract the characteristics of the non-stationary signals; therefore, used in this paper to analyze the cracked beam vibrational signals. The wavelet power spectrum (WPS) calculated for different beam health conditions and for different locations of the vibration sensors. The results reveal the effectiveness of the wavelet power spectrum over the traditional FFT spectrum to identify the existence of the crack in the wind turbine blade. Moreover, a mathematical model, which correlates the variation of the blade natural frequencies with the crack severity and location, is presented and validated with an experimental modal analysis data.