Multi-objective Optimization and Multi-criteria Decision Making in sls.pdf

Multi-objective Optimization and Multi-criteria Decision Making
in SLS
Nikhil Padhye, Subodh Kalia and Kalyanmoy Deb
npdhye@, ******@, ******@
Department of Mechanical Engineering
Indian Institute of Technology Kanpur, Kanpur-208016, ., India
November 22, 2009
Abstract
This paper proposes an integrated approach to arrive at optimal build orientations, simultaneously
minimizing surface roughness ‘Ra’ and build time ‘T’, for object manufacturing in SLS process. The
optimization task is carried out by two popularly known multi-objective evolutionary optimizers -
NSGA-II (non-dominated sorting ic algorithm) and MOPSO (multi-objective particle swarm op-
timizer). The parison of these optimizers and an approximation of Pareto-optimal
front is done using two statistically significant performance measures. Three proposals addressing the
task of decision making, . selecting one solution in presence of multiple trade-off solutions, are
made to facilitate the designer. A hill climbing local search procedure is also proposed to further refine
the solutions obtained by evolutionary optimizers. The overall procedure is integerated into a MORPE
- Multi-objective Rapid Prototyping Engine. Several sample objects are considered for experimenta-
tion to demonstrate the working of MORPE. A careful study of optimal build directions for several
components indicates a trend, providing insight into the SLS processes which can be regarded highly
useful for various practical RP applications.
Keywords: Multi-objective Optimization, Decision Making, ic Algorithms, Particle Swarm Op-
timization and SLS.
1 Introduction
Rapid prototyping (RP) or layered manufacturing refers to processes in which ponent is fabricated by
layer-by-layer deposition of material from puter asisted design models. RP is playing an important
role in reducing the time required for new product development and lowering development costs, thus
panies are rea