Towards optimal ship design and valuable knowledge discovery under uncertain conditions
Abstract
Ship design is a complex engineering activity which requires a multidisciplinary consideration in arriving at design objectives and constraints. An optimal design of such problems require a multi-objective optimization method that is capable of finding multiple trade-off solutions, not only to choose a preferred solution for implementation, but also to have a deeper understanding of the interactions among design variables. In this paper, we consider two ship design models involving uncertainties in design variables, and demonstrate the usefulness of an evolutionary multiobjective optimization (EMO) method and subsequent data analysis procedures in arriving at valuable design principles that enhance the knowledge of a designer. The study is pedagogical yet provide key insights of ship design issues and importantly outlines the systematic procedure for applying the technology to other more complex design problems.
Authors: Kalyanmoy Deb, Zhichao Lu, Chris B. McKesson, Cherie Courseault Trumbach, Larry DeCan
Published in: IEEE Congress on Evolutionary Computation (CEC) (2015)