Affine LPV-modeling for the ADDSAFE benchmark

• The proposed method allows us to generate LPV models with large validity regions.
• The sparsity of the LPV models helps to generate LFT representations of low order.
• Large validity regions and sparsity of LPV models reduce the computational effort for LPV/LFT based analysis and synthesis methods.

General approaches are presented for approximating the nonlinear dynamics of a commercial aircraft model by affine linear parameter varying (LPV) models that are used for the design of fault detection and diagnosis (FDD) systems. One part of the paper deals with the approximation of the nonlinear actuator dynamics, where a simple analytic model is derived that accurately describes a given nonlinear, (partly) black-box model in the whole flight envelope. The second part presents a new two step approach for generating an affine LPV model for the nonlinear aircraft dynamics. Starting from a given trim-point in the flight envelope, the goal of the first step is to maximize the size of the region around this trim point, for which the simple affine LPV description is still valid. The second step then tries to further simplify the affine LPV model, which helps to reduce both the computational burden for FDD synthesis methods and the order of the linear fractional representations (LFRs) that are generated from these LPV models.