Summary
International Symposium on Nonlinear Theory and its Applications
2008
Session Number:B3L-B
Session:
Number:B3L-B3
The Quest for a Shady Place: A guide (using shadowing filters for state estimation)
Thomas Stemler, Kevin Judd,
pp.-
Publication Date:2008/9/7
Online ISSN:2188-5079
DOI:10.34385/proc.42.B3L-B3
PDF download (235.5KB)
Summary:
Given a time series of a physical system and some model of it, the quality of forecasts rely on a satisfactory model and state estimation. Small alterations in the initial state can result in large forecast errors, even when the model is perfect. We report on a promising new tool for state estimation, the shadowing filters. We compare its performance with the classical extended Kalman filter and demonstrate that it is superior. To ensure successful implementation of a shadowing filter in an arbitrary model we advise using several tests. These tests cover many important aspects of a real-world applications, such as: data requirements, identification of errors, and the cost performance of different update routines for sequential state estimation. This talk reports on the superior performance of shadowing filters and provides a “how to” for implementing them.