Calculating wind variances for better performance
Changes in wind speed and direction can play havoc on wind turbines if left unchecked. Not only can they adversely affect their optimal performance, they can also impact their longevity. If the wind is too strong, for example, the turbines have to be shut down. Turbines are carefully balanced instruments, requiring constant monitoring to ensure optimal conditions and these algorithms provide suitable means by which these can be achieved. The algorithms are capable of functioning both on a predictive and on an immediate response basis. They do so under two main headings; medium and low frequency. Medium frequency operates at the 30-32Hz range and analyses the power signal by means of wavelets and Fast Fourier transforms (FFTs). With the use of these algorithms the generator shaft misalignment could be detected by simply looking at the generator slip frequency. More specifically, fine details in the electric power signal provide valuable information on changes in the generator flux field during periods of increased vibrations. Low frequency analysis uses data from Supervisory Control and Data Acquisition (SCADA) systems to assess temperature signals over time and improve the quality of failure detection. This determines any possible faults that might occur in the bearings, pitch, yaw, anemometer and controllers. The research was intended to benefit project partners in improving their ability to monitor offshore wind farms remotely and is directly related to research looking into automated surveillance systems beyond the scope of the CONMOW project. On the other hand, the developed algorithms are readily applicable to wind-driven generators and can improve their ability to function under a wide range of wind conditions by contributing to pro-active and predictive maintenance efforts.
