Incorporating linear local models in Gaussian process model
Abstract
Identification of nonlinear dynamic systems from experimental datacan be difficult when, as often happens, more data are availablearound equilibrium points and only sparse data are available farfrom those points. The probabilistic Gaussian Process model has already proved to modelsuch systems efficiently. The purpose of this paper is to show howone can relatively easily combine measured data and linear localmodels in this model. Also, using previous results, we showhow uncertainty can be propagated through such models whenpredicting ahead in time in an iterative manner. The approach isillustrated with a simple numerical example.
