Probabilistic clustering of wind energy conversion systems using classification models

Abstract

This research intends to give insights on the pattern aggregation of wind energy conversion systems technologies through identification of homogeneous groups within a set of wind farms installed in Portugal. Pattern aggregation is performed using Hierarchical Cluster Analysis followed by Discriminant Analysis, in order to validate the results produced by the first one. The clustering support matrix uses three independent variables: installed capacity, net production and capacity factor, in a per year basis. Cluster labelling allows the identification of two homogenous groups of wind farms, whose main attributes are based on the technological conversion system trend: (1) asynchronous generator based technology and (2) direct driven synchronous generator based technology, with higher capacity factors.