Non linear system identification : a state -space approach

In this thesis, new system identication methods are presented for three particular types of nonlinear systems: linear parameter-varying state-space systems, bilinear state-space systems, and local linear state-space systems. Although most work on nonlinear system identication deals with nonlinear input-output descriptions, this thesis focuses on state-space descriptions. State-space systems are considered, because they are especially suitable for dealing with multiple inputs and outputs, and they usually require less parameters to describe a system than input-output descriptions do. Equally important, the starting point of many nonlinear control methods is a state-space model of the system to be controlled.