Graduate Thesis Or Dissertation
 

Parameter estimation in system modeling using man-machine interaction and computer learning

Public Deposited

Downloadable Content

Download PDF
Request Accessible Version
https://ir.library.oregonstate.edu/concern/graduate_thesis_or_dissertations/tm70mz48v

Descriptions

Attribute NameValues
Creator
Abstract
  • In certain cases in system modeling, parameter search problems are complicated by the availability of scanty and corrupt physical system data and a large number of model parameters. An algorithm using man-machine interaction is presented to attack this type of parameter search problem with the objective of efficiency in the parameter search. The man-machine interaction allows model response evaluation by a man by providing the man with a convenient display of the data and the model responses. The large number of parameters in the system model contributes to a slow parameter search. The algorithm presented uses computer learning to improve the parameter search efficiency. The algorithm is divided into a learning phase and a learned phase. In the learning phase the man teaches the computer his preferences for the model responses. Then in the learned phase the computer proceeds with the parameter search independently of the man by means of its experience acquired during the learning phase. After a period of learned model response evaluation by independent action of the computer in the learned phase the control of the parameter search process is returned to the man. This terminates one iteration of the algorithm. The problem of extracting features from model responses to train the computer is studied. A method to compare different feature extraction operations using entropy is presented. The algorithm is tested on an example oceanographic problem. It is found that the success of the computer learning technique depends on the step size of the model parameter changes during the learned (computer directed) phase of operation and on the size of a learned region in feature space (a quantity calculated for response classification in the learned phase). Also, the algorithm is more efficient when the parameters are not too close to their optimum.
Resource Type
Date Available
Date Issued
Degree Level
Degree Name
Degree Field
Degree Grantor
Commencement Year
Advisor
Academic Affiliation
Non-Academic Affiliation
Subject
Rights Statement
Publisher
Peer Reviewed
Language
Digitization Specifications
  • File scanned at 300 ppi (Monochrome) using ScandAll PRO 1.8.1 on a Fi-6670 in PDF format. CVista PdfCompressor 4.0 was used for pdf compression and textual OCR.
Replaces

Relationships

Parents:

This work has no parents.

In Collection:

Items

Thumbnail Title Date Uploaded Visibility Actions
file details WagnerJohnA1970.pdf 2017-08-17 Public Download