An Estimation of Nadir Objective Vector using a Hybrid Evolutionary-Cum-Local-Search Procedure
2009 (English)Report (Other academic)
A nadir objective vector is constructed from the worstPareto-optimal objective values in a multi-objectiveoptimization problem and is an important entity tocompute because of its significance in estimating therange of objective values in the Pareto-optimal frontand also in executing a number of interactive multi-objective optimization techniques. Along with theideal objective vector, it is also needed for the purposeof normalizing different objectives, so as to facilitatea comparison and agglomeration of the objectives.However, the task of estimating the nadir objectivevector necessitates information about the completePareto-optimal front and has been reported to be adifficult task, and importantly an unsolved and openresearch issue. In this paper, we propose certain mod-ifications to an existing evolutionary multi-objectiveoptimization procedure to focus its search towardsthe extreme objective values and combine it with areference-point based local search approach to con-stitute a couple of hybrid procedures for a reliableestimation of the nadir objective vector. With upto 20-objective optimization test problems and on athree-objective engineering design optimization prob-lem, one of the proposed procedures is found to becapable of finding the nadir objective vector reliably.The study clearly shows the significance of an evolu-tionary computing based search procedure in assist-ing to solve an age-old important task in the field ofmulti-objective optimization.
Place, publisher, year, edition, pages
Helsinki School of Economics Print, 2009.
Nadir point, multi-objective opti- mization, non-dominated sorting GA, evolutionary multi-objective optimization (EMO), multiple objectives, hybrid procedure, ideal point, Pareto optimality.
Economics and Business
IdentifiersURN: urn:nbn:se:kth:diva-82970OAI: oai:DiVA.org:kth-82970DiVA: diva2:498642
QC 201203062012-02-122012-02-122012-03-06Bibliographically approved