Computation of the observer gain for extended Luenberger observers using automatic differentiation

K Röbenack - IMA Journal of Mathematical Control and …, 2004 - ieeexplore.ieee.org
IMA Journal of Mathematical Control and Information, 2004ieeexplore.ieee.org
For a given nonlinear system, the extended Luenberger observer provides nearly exact error
dynamics. In contrast to the normal form observer, the extended Luenberger observer exists
even if the associated integrability condition is violated. Up to now, Lie derivatives and Lie
brackets required by the design procedure have been computed symbolically. Even for
systems with moderate size and complexity, one usually obtains extremely large
expressions for the observer gain. The design of an extended Luenberger observer based …
For a given nonlinear system, the extended Luenberger observer provides nearly exact error dynamics. In contrast to the normal form observer, the extended Luenberger observer exists even if the associated integrability condition is violated. Up to now, Lie derivatives and Lie brackets required by the design procedure have been computed symbolically. Even for systems with moderate size and complexity, one usually obtains extremely large expressions for the observer gain. The design of an extended Luenberger observer based on symbolic differentiation is not feasible for complicated or large‐scale systems. In this paper we discuss a new approach to compute the observer gain. Our approach is based on a computation method for derivatives called automatic differentiation. In contrast to numeric differentiation by means of divided differences, automatic differentiation incurs no truncation errors.
ieeexplore.ieee.org
以上显示的是最相近的搜索结果。 查看全部搜索结果