A novel predictive maintenance method for wind turbines based on wavelets transforms

Wind turbines maintenance management has changed over the past to achieve time and cost reductions, and to increase the reliability, availability, maintainability and safety (RAMS) of wind turbines. One of the most common techniques is the Supervisory Control and Data Acquisition (SCADA) system applied to wind turbines in order to guarantee correct levels of RAMS. This paper describes a novel fault detection and diagnosis method based on a pattern recognition approach for sound and vibration signals. The main conclusion in this study has been to establish a pattern recognition between vibration and sound for fault detection and diagnosis. The sound signal has been analyzed by wavelet decomposition technique. The signal is decompounded into a set of wavelets coefficients and a percentage of energy is associated to each one. Vibration patterns for the Fast Fourier Transform, combined with sound analysis using the theory of wavelets and their energy states, allow the study of the behavior of signals from a different point of view. This paper has carried for any type of maintenance that involves the existence of any rolling element to find potential faults such as misalignment, looseness, unbalances, etc.