The Identification of Nonlinear Dynamical Systems from Input-Output Data.

Abstract

The approaches used in the identification of nonlinear systems from input-output data have normally assumed that the system is almost completely unknown, or that it is completely known with the exception of fixed number of unknown parameters. In this report it is assumed that the input-output relations for the nonlinear dynamical system under study are known with the exception of an unknown function which occurs in this representation. From a finite length of input data and noise corrupted output data techniques are presented which yield a maximum likelihood estimate of an approximation to the unknown function. Computer algorithms based on gradient and Newton-Raphson procedures are developed, and an example is presented. (Author)

Document Details

Document Type
Technical Report
Publication Date
Jul 01, 1969
Accession Number
AD0858698

Entities

People

  • Fletcher R. Phillips

Organizations

  • University of California, Los Angeles

Tags

DTIC Thesaurus Topics

  • Algorithms
  • Computers
  • Computing Devices
  • Identification
  • Nonlinear Systems

Readers

  • Approximation Theory.
  • Neural Network Machine Learning.
  • Systems Analysis and Design