Approximations and Implementations of Nonlinear Filtering Schemes

摘要

The objective of this program is to develop approximation models that may be used for the implementation of nonlinear filtering schemes. The approach is based on two major concepts. The first is concerned with the approximation of nonlinear models for air-to-air missile scenarios by a Markov linear approximate model. The second concept is concerned with models for the errors due to the finite word-length implementation of the filters. In addition, the project is involved with the study of hybird system models that result when nonlinear or linear models such as these are quantized and are subject to switches among a set of linear models. The switching can either be deterministic or stochastic and may involve fast and slow dynamics.