Tchebycheff Method-based Evolutionary Algorithm for Multiobjective Optimization
No Thumbnail Available
Files
Date
2003-09-05
Authors
Journal Title
Series/Report No.
Journal ISSN
Volume Title
Publisher
Abstract
In the operations research literature, the Tchebycheff method has been demonstrated to be a useful approach for exploring the non-dominated solutions for multiobjective optimization problems. While this method has been investigated with mathematical programming-based solution approaches, its application with modern heuristic search procedures is lacking. As heuristic search procedures continue to show promise as practical solution approaches for realistic engineering problems typically with multiple design objectives, the need for their applications in multiobjective optimization is becoming increasingly important. This paper investigates a new evolutionary algorithm-based multiobjective optimization procedure that builds upon the Tchebycheff method. By embedding a beneficial seeding approach, the efficiency of the algorithm is expectedly enhanced. This Tchebycheff Method-based Evolutionary Algorithm (TMEA) is tested and evaluated using a suite of 2-objective test problems, representing a range of complexities in the decision space as well as in the objective space. The performance of TMEA with those of other multiobjective evolutionary algorithms are compared using several performance metrics that are reported in the literature. For the problems considered in this paper, TMEA performs relatively well in generating non-dominated solutions that are close to the known Pareto set and are well distributed in the non-inferior space.
Description
Keywords
Multiobjective Optimization, Evolutionary computation
Citation
Degree
MS
Discipline
Civil Engineering