Premium
A multicriteria evolutionary algorithm for mechanical design optimization with expert rules
Author(s) -
Coelho R. Filomeno,
Bouillard Ph.
Publication year - 2004
Publication title -
international journal for numerical methods in engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.421
H-Index - 168
eISSN - 1097-0207
pISSN - 0029-5981
DOI - 10.1002/nme.1198
Subject(s) - parameterized complexity , context (archaeology) , computer science , inference , mathematical optimization , process (computing) , evolutionary algorithm , algorithm , mathematics , artificial intelligence , paleontology , biology , operating system
This paper addresses the problem of optimizing mechanical components during the first stage of the design process. While a previous study focused on parameterized designs with fixed configurations—which led to the development of the PAMUC ( Preferences Applied to Multiobjectivity and Constraints ) method, to tackle constraints and preferences in evolutionary algorithms (EAs)—, the models to be considered in this work are enriched by the presence of topological variables. In this context, in order to create optimal but also realistic designs, i.e. fulfilling not only technical requirements but also technological constraints (more naturally expressed in terms of rules), a novel approach is proposed: PAMUC II. It consists in integrating an inference engine within the EA to repair the individuals violating the user‐defined rules. PAMUC II is tested on mechanical benchmarks, and provides very satisfactory results in comparison with a weighted sum method with penalization to deal with the constraints. Copyright © 2004 John Wiley & Sons, Ltd.