Date of Award

Spring 1967

Document Type

Thesis - Restricted

Degree Name

Master of Science (MS)

Department

Electrical and Computer Engineering

First Advisor

Blank, Gary L.

Second Advisor

Ishii, Thomas K.

Abstract

This thesis is concerned with the modern control theory concept of observability and its relationship with Kalman optimal and suboptimal linear estimation theory. The interrelation between the accuracy of suboptimization and the degree of observability for dynamical systems is demonstrated with a second order model. The accuracy of suboptimization is determined by comparing the diagonal elements in the covariance matrix of the optimal estimation error with the corresponding diagonal elements in the covariance matrix of the suboptimal estimation error . The percentage err or between these elements is plotted for a parameter variation in the model. The degree of observability is measured by the magnitude of a "loss function." This loss function is plotted for the same variation in the model parameter. It is verified in the thesis that if the two state variables of the second model are not observable, then the error between optimal and suboptimal linear estimation is a minimum.

Share

COinS

Restricted Access Item

Having trouble?