摘要：State-space identification of Linear Parameter-Varying (LPV) models using local data still represents a significant challenge as interpolation of local state-space models suffers from the well-known state-basis coherence problem. Recently, various behavioral-type of interpolation methods, in which a global LPV model is constructed based on matching its input-output behavior with the local models, have been introduced to overcome this issue. However, these methods suffer from high computational complexity and stability restrictions of the local models. In this paper, a novel method is introduced that is based on direct local-matrix norm matching, which, contrary to previous works, does not suffer from basis incoherence of the local models and neither requires their stability. Although these properties are highly desirable, a simulation study on an e-Nose sensor system indicates that the method in its current form is not robust to noise and therefore future research is needed to apply the presented concept in a realistic system identification setting.
