Machine learning based model linearization of a wind turbine for power regulation

ABSTRACT

Wind turbine systems exhibit highly nonlinear dynamics influenced by the aerodynamic torque induced in the wind turbine blades and thrust force on the turbine structure due to the wind flow. This paper presents a system identification approach to approximate the nonlinear wind turbine model. A clustering-based piecewise affine system identification technique is utilized to construct an affine multiple-model that is valid for the power regulation region of a wind turbine. A comprehensive study is performed to validate the accuracy and performance of the developed model. The piecewise affine model identified in this paper can be widely used for advanced control systems design and the security assessment of the power grid.

KEYWORDS:

• Piecewise affine model
• wind turbine
• clustering-based identification
• classification
• system identification
• nonlinear systems