Statistical Properties of a Least Squares Identification Algorithm Used in Modeling Human Operator Performance.

Abstract

Three methods for identifying the parameters of a linear, autonomous, discrete time single input-single output system are studied. These are a least squares method, known as the output error method, a maximum likelihood method due to Kashyap, and a linear least squares method due to Levin. On the basis of extensive simulation studies, it is concluded that the output error method is the best of the three algorithms. It is shown that if the plant noise and observation noise are independent sequences of independent and identically distributed random variables with finite second moments, then the least squares estimator is consistent. In addition, if the noises have finite third moments, the estimator is asymptotically normal. The output error method is successfully applied to some tracking data.

Document Details

Document Type
Technical Report
Publication Date
May 01, 1975
Accession Number
ADA012705

Entities

People

  • Donald O. Norris
  • Larry E. Snyder

Organizations

  • Ohio University

Tags

DTIC Thesaurus Topics

  • Algorithms
  • Estimators
  • Identification
  • Least Squares Method
  • Mathematics
  • Observation
  • Optimal Estimators
  • Random Variables
  • Sequences
  • Simulations
  • Statistical Algorithms

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Finite Element Method (FEM) for solving Partial Differential Equations (PDEs)
  • Statistical inference.