Optimum Adaptive Piecewise Linearization: An Estimation Approach in Wind Power

Abstract

This paper introduces an effective piecewise linearization technique to obtain an estimation of nonlinear models when their input-output domains include multidimensional operating points. The algorithm of a forward adaptive approach is introduced to identify the effective operating points for model linearization and adjust their domains for the maximum coverage and the minimum model linearization error. The technique obtains a minimum number of linearized models and the continuity of their domains. The algorithm also yields global minimum model linearization error. The introduced algorithm is formulated for a wind power transfer system for a 2-D set of input domains. The linearization error can be arbitrarily minimized in exchange for a higher number of models. The results demonstrate a significant improvement in the linearization of nonlinear models.