An evolutionary algorithm to generate alternatives (EAGA) for engineering optimization problems

An evolutionary algorithm to generate alternatives (EAGA) for engineering optimization problems

0.00 Avg rating0 Votes
Article ID: iaor20051528
Country: United Kingdom
Volume: 36
Issue: 5
Start Page Number: 539
End Page Number: 553
Publication Date: Oct 2004
Journal: Engineering Optimization
Authors: ,
Keywords: heuristics, optimization, stochastic processes
Abstract:

Typically for a real optimization problem, the optimal solution to a mathematical model of that real problem may not always be the ‘best’ solution when considering unmodeled or unquantified objectives during decision-making. Formal approaches to explore efficiently for good but maximally different alternative solutions have been established in the operations research literature, and have been shown to be valuable in identifying solutions that perform expectedly well with respect to modeled and unmodeled objectives. While the use of evolutionary algorithms (EAs) to solve real engineering optimization problems is becoming increasingly common, systematic alternatives-generation capabilities are not fully extended for EAs. This paper presents a new EA-based approach to generate alternatives (EAGA), and illustrates its applicability via two test problems. A realistic airline route network design problem was also solved and analyzed successfully using EAGA. The EAGA promises to be a flexible procedure for exploring alternative solutions that could assist when making decisions for real engineering optimization problems riddled with unmodeled or unquantified issues.

Reviews

Required fields are marked *. Your email address will not be published.