Data-Driven Wind Turbine Wake Decomposition and Evolution Analysis: A Frequency Characteristics-based Clustering Procedure

With the research in wind turbine wake deepening, frequency characteristics are gaining more attention. Data-driven methods are usually adopted to analyze the wake performance. In this paper, a frequency characteristics-based clustering procedure is proposed to decompose the wind turbine wake into clusters, of which the wake frequency-domain performance and dynamic wake evolution process can be analyzed. The proposed procedure first calculates the frequency amplitudes of the measured velocity time-series on every measurement point in space, then decomposes the wake according to the amplitudes using a clustering algorithm. Take the centroid in each cluster as an example, dominant frequency characteristics are analyzed, which show the wake meandering frequency while help the understanding of the dynamic wake evolution.