Local E-optimality Conditions for Trajectory Design to Estimate Parameters in Nonlinear Systems

TitleLocal E-optimality Conditions for Trajectory Design to Estimate Parameters in Nonlinear Systems
Publication TypeConference Paper
Year of Publication2014
AuthorsWilson, A. D., and T. D. Murphey
Conference NameAmerican Control Conference (ACC)
Pages443-450
Abstract

This paper develops an optimization method to synthesize trajectories for use in the identification of system parameters. Using widely studied techniques to compute Fisher information based on observations of nonlinear dynamical systems, an infinite-dimensional, projection-based optimization algorithm is formulated to optimize the system trajectory using eigenvalues of the Fisher information matrix as the cost metric. An example of a cart-pendulum simulation demonstrates a significant increase in the Fisher information using the optimized trajectory with decreased parameter variances shown through Monte-Carlo tests and computation of the Cramer-Rao lower bound.

DOI10.1109/ACC.2014.6858649
Publication PDF: