Date of Award

Spring 2006

Document Type

Thesis - Restricted

Degree Name

Master of Science (MS)

Department

Electrical and Computer Engineering

First Advisor

Arkadan, Abdul-Rahman A.

Second Advisor

Yaz, Edwin E.

Third Advisor

Johnson, Michael

Abstract

A Recursive Fuzzy Logic Classifier, R-FL-C model is presented in this work for the design optimization of Permanent Magnet Synchronous generators. The aim of the optimization process is to minimize the current level of harmonics, total losses and maximizing the developed power. The optimization technique uses the data obtained offline from the Finite Element State Space, FE-SS model of aerospace PM synchronous generators. Moreover, the R-FL-C consists of two modules. The first module is a Classification Module that utilizes a Fuzzy Logic Classifier to project the two dimensional discrete search space into one dimensional discrete search space. It also analyzes the statistics such as the range and cluster of the net cost function in order to create the knowledge to perform the projection. The second is a Recursive-Fuzzy Logic module which is recursively utilized to search for the optimal solution by minimizing the resulting one dimensional cost function.

Share

COinS

Restricted Access Item

Having trouble?