Cluster-based probabilistic structure dynamical model of wind turbine wake

Abstract

For complex flow systems like the one of the wind turbine wakes, which include a range of interacting turbulent scales, there is the potential to reduce the high dimensionality of the problem to low-rank approximations. Unsupervised cluster analysis based on the proper orthogonal decomposition is used here to identify the coherent structure and transition dynamics of wind turbine wake. Through the clustering approach, the nonlinear dynamics of the turbine wake is presented in a linear framework. The features of the fluctuating velocity are grouped based on similarity and presented as the centroids of the defining clusters. Determined from probability distribution of the transition, the dynamical system identifies the features of the wakes and the inherent dynamics of the flow.