MCDM'08 - paper no. 1


 

Back to MCDM'08 contents
 

MCDM with relative importance of criteria: application to configuration design of vehicles

Vincent Y. Blouin, Brian J. Hunt, Margaret M. Wiecek

Abstract:

In this article, a preference model of relative importance of criteria is presented and applied to a vehicle configuration design problem formulated as a multi-objective program. The model is based on convex cones and extends the classical notion of Pareto optimality. Preferences quantifies as allowable tradeoffs are elicited from the decision maker and used for the construction of a new preference. In effect, the Pareto set is reduced which facilitates the process of choosing a final preferred solution.
Configuration design of mechanical systems corresponds to finding the placement of a set of components such that performance criteria are optimized while satisfying design constraints. The presented configuration design problem involves designing a midsize truck for optimum vehicle dynamic behavior, survivability, and maintainability in the presence of decision maker's preferences that are included a posteriori. A set of Pareto solutions is first generated with a multi-objective genetic algorithm, and the Pareto solutions are screened according to the preferences quantified as allowable tradeoffs. The model extracts preferred designs from the Pareto set producing a short list of "strong" or "privileged" designs, which is a useful feature when preferences are unknown.

Keywords:

Multi-criteria decision-making, preferences, Pareto optimality, relative importance, tradeoffs, Pareto set, configuration design, vehicle design

Reference index:

Vincent Y. Blouin, Brian J. Hunt, Margaret M. Wiecek, (2009), MCDM with relative importance of criteria: application to configuration design of vehicles, Multiple Criteria Decision Making (4), pp. 11-40

Full text:

download

Scopus citations in 1 paper(s):
  1. Faulkenberg, S. L., & Wiecek, M. M. (2011, 4). Generating equidistant representations in biobjective programming. Computational Optimization and Applications, 51, 1173-1210. doi:10.1007/s10589-011-9403-5