Studies on enhanced operator-oriented genetic algorithms
Abstract
Genetic Algorithms (GAs) are robust search algorithms used in a variety of applications, including engineering problems. GAs provide an alternative to traditional optimization techniques by using directed random searches to locate optimal solutions for complex problems. Until recently, most GA applications dealt with the optimization of static combinatorial problems. Solutions to these problems are typically mapped into a representation domain where specific forms of GA search operators are used to search a space defined by the order independent solution structure of fixed length. For this dissertation, a new order-based GA, denoted as Enhanced Operator Oriented Genetic Algorithm (EOOGA), is proposed to solve dynamic problems where solutions are expressed by a varying length sequence of decision vectors. The EOOGA transforms an initial solution state to a final solution state while satisfying constraints and objectives. A new procedure is incorporated into the EOOGA that solves problems in incremental steps, by under-specifying solutions and cascading successive searches. In our approach, large complex problems become more manageable, and solutions become more attainable. An experimental investigation via computer simulations is done to examine different generation strategies that enhance EOOGA performance.
This paper has been withdrawn.