Tuning the Discrete Wavelet Transform for Power Smoothing of Wind Turbines

Abstract

Abstract In the study, an extended sensitivity analysis is presented, which was aimed at properly tuning the parameters of an algorithm based on the Discrete Wavelet Transform (DWT) for use in power smoothing of utility-scale wind turbines coupled with batteries. More specifically, a twofold implementation is proposed, so that the proposed algorithm can operate efficiently both in real time as a control system and using historical data for the preliminary sizing of the storage system. In particular, this study addresses the correct setting of the main parameters of DTW, i. e. the level of decomposition and the mother wavelet family that generates the multi-resolution analysis (MRA). Based on real wind data of an onshore site, the following wavelet families have been analyzed: Daubechies, Coiflet, Symmlet, Biorthogonal and Reverse Biorthogonal. It is shown that, as the severe wind fluctuations that need to be smoothed are a quite sudden phenomena, in which usually the wind speed increases and then decreases quickly, all the wavelet families having a centered peak show good performance. On the other hand, it is highlighted that, once the correct choice of the mother wavelet is made, neither increasing the decomposition level nor making it adjustable in time, brings significant benefits. Finally, the discussed hypothesis has been assessed in combination with the proposed technique to extend the wavelet for online control using data mirroring, corroborating the suitability of the method for use in wind energy applications.