Date of Award
Summer 1996
Document Type
Dissertation - Restricted
Degree Name
Doctor of Philosophy (PhD)
Department
Electrical and Computer Engineering
First Advisor
Feng, Xin
Second Advisor
Belfore, Lee
Third Advisor
Heinen, James
Abstract
This dissertation is devoted to the study and development of a modified order-based Genetic Algorithm called the Enhanced Operator Oriented Genetic Algorithm (EOOGA). For problems which require solutions to be expressed in terms of a sequence of operations, EOOGA provides a unique framework from which solutions can be directly represented and evolved. In addition, a new search strategy is developed to improve EOOGA's performance in large and complex domains. This dissertation also examines different generational strategies to find the one most suited to maximize EOOGA performance. EOOGA performance is evaluated on two types of EOOGA applications. The first application entails finding a sequence of operations to solve a combinatorial problem; e.g., 15 puzzle problem and Rubik's cube. The second application involves obtaining near optimal control policies for two powered flight problems and for the steering of a truck-and-trailer problem.