Vibration-based tools for the optimisation of large-scale industrial wind turbine devices

Wind turbine (WT) maintenance management must be in continuous improvement to develop reliability, availability, maintainability and safety (RAMS) programmes and to achieve time and cost reductions in large-scale industrial wind turbines. The optimisation of the operation reliability involves supervisory control and data acquisition to guarantee the correct levels of RAMS. A fault detection and diagnosis methodology (FDD) is proposed for the mechanical devices of a WT. The method applies the wavelet and Fourier analysis to vibration signals. The signals collected contain information on failures found in the gearbox-generator set. The information is initially tested by the fast Fourier transform (FFT) to ensure its accuracy. Then, a pattern is created based on energies that relate each failure to different frequency bands. This pattern uses the wavelet transform as the main technique. A number of turbines of the same type were instrumented in the same wind farm. The data collected from the individual turbines was fused together and analysed in order to determine the overall performance. It is expected that data fusion allows for a significant improvement, since the information gained from various condition monitoring systems can be enhanced. The paper will also focus on the application of dependable, embedded computer systems for a reliable implementation.