Modeling wind-turbine power curve: A data partitioning and mining approach

Abstract

Model of a power curve allows to analyze performance of a wind turbine and compare it with other turbines. An approach based on centers of data partitions and data mining is proposed to construct such a model. Wind speed range is partitioned into intervals for which centers are computed. The centers are regarded as representative samples in modeling. A support vector machine algorithm is used to build a power curve model. Computational results have demonstrated that the model reflects dynamic properties of a power curve. In addition it is accurate and efficient to generate. The model accuracy has been tested with industrial wind energy data.