Accession Number:

ADA008729

Title:

Comparison of Four On-Line Identification Methods for Multivariable Systems,

Descriptive Note:

Corporate Author:

CALIFORNIA UNIV SANTA BARBARA DEPT OF ELECTRICAL ENGINEERING

Personal Author(s):

• Nelson,Lawrence W.
• Stear,Edwin B.

Report Date:

1974-01-01

Pagination or Media Count:

8.0

Abstract:

This paper presents the results of a comparison of four methods for on-line identification of discrete time, linear, time invariant, multivariable systems. The methods chosen for comparison are, the extended Kalman filter, instrumental variables method, extended least squares and stochastic approximations. These methods are applied to both a multivariable system as well as a single input single output system. The computational properties of the methods are compared as well as their convergence properties. Various trade offs between the methods are pointed out.

Descriptors:

• *CONTROL THEORY
• LINEAR SYSTEMS
• STOCHASTIC PROCESSES
• KALMAN FILTERING
• IDENTIFICATION
• LEAST SQUARES METHOD

Subject Categories:

• Statistics and Probability

Distribution Statement:

APPROVED FOR PUBLIC RELEASE