Similarity based local model approach for nonlinear modelling

This contribution considers a new method for modelling nonlinear dynamical systems with blended multiple model structure. By approximating the nonlinear system sufficiently accurate, the model can be used as a basis for controller design. The local similarity between the linear local models and the nonlinear system is used to build the model structure. The obtained approximating system and the controller can be regarded as a Sugeno-Takagi fuzzy system. The method is outlined using simulated rotor dynamics of a wind turbine system.