Solving constrained multi-objective problems by objective space analysis

In this paper a new approach to solve constrained multi-objective problems by way of evolutionary multi-objective optimization is introduced. In contrast to former evolutionary approaches, which amalgamate objective space dominance relations with feasibility of solutions considered in the design spaces, the hereby suggested approach relies solely on objective space based analysis. It is shown in this paper that considering the violation of constraints within the design space is problematic as it may lead to misleading conclusions. Moreover, the current approach is inherently capable of dealing with constraints that are imposed directly in the objective space. Categories and Subject Descriptors I.2.8 [Artificial Intelligence]: Problem Solving - Heuristic Methods J.6.1 [Computer-aided Engineering]: Computer-aided Design General Terms Algorithms, Design. Keywords Multi-objective, Constraints, Reliability.

• Type: Conference
• Year: 2008
• Where: GECCO
• Authors: Gideon Avigad, Carlos A. Coello Coello
• Keywords: Design Space, GECCO 2008, Objective Space, Objective Space Dominance, Optimization