Abstract

This paper proposes a Bayesian framework based on particle filters for online fatigue damage diagnosis and prognosis for wind turbine blades (WTBs). The framework integrates theoretical and practical aspects with the purpose of developing a robust monitoring tool. Besides, a damage indicator based on identified modal frequencies of the WTB is defined to quantify the degree of damage to the monitored blade. Furthermore, feature extraction techniques on vibration signals are considered to obtain the online observations and inputs needed by the Bayesian framework. Experimental data collected from a fatigue test performed on a WTB was used to validate the proposed methodology. The results obtained by the damage diagnosis algorithm show the great potential of the Bayesian processor for damage estimation of the monitored WTB and uncertainty quantification related to the estimated variable. According to the damage prognosis results, the proposed algorithm could generate suitable long-term predictions of the damage indicator so as to estimate the time-of-failure (ToF) probability mass function (PMF) for the monitored WTB. Consequently, the computed ToF-PMF was able to include the experimental ToF within its 95% confidence interval, demonstrating the accuracy of the Bayesian approach.