Multi-objective optimizations of mechanical characteristics of objects in computer aided SIS manufacturing process using empirical PSO algorithm
Implementation of the “Selective Inhibition type of Sintering (SIS)” method in the field of additive manufacturing, which develops the components by building layers, depends upon the proper selection of their optimum input attributes. Components manufacturing as per anticipated size is accomplishe...
Main Authors: | , |
---|---|
Format: | Article |
Language: | English |
Published: |
ICI Publishing House
2020-09-01
|
Series: | Revista Română de Informatică și Automatică |
Subjects: | |
Online Access: | https://rria.ici.ro/wp-content/uploads/2020/09/art._Yunus_Alsoufi.pdf |
Summary: | Implementation of the “Selective Inhibition type of Sintering (SIS)” method in the field of additive
manufacturing, which develops the components by building layers, depends upon the proper selection of their
optimum input attributes. Components manufacturing as per anticipated size is accomplished by the
optimization of factors, namely, heater power (H), layer height (L), the feed rate of the heater (F) and roller
(R), and support’s temperature (S) using “Multi-Response Particle Swarm Optimization (MRPSO)”. Analysis
of variance was employed to substantiate the competence of the established models. Trials have shown that
mechanical properties of manufactured components are characterized by Design of experiments in a direct
relationship with F and R, but inversely related to H and L. Finally, unique MRPSO algorithm has been
employed for parallel optimization of multiple outputs. Introduced mutation functions (of genetic algorithm)
into MRPSO algorithm to prevent early convergence. Optimal-solutions of Pareto-front achieved by MRPSO
were graded by the compound grades (resulting from maximum deviation theory) to increase preciseness in
making the decisions. The MRPSO analysis offered valuable information for monitoring the factors to
improve the accuracy of SIS components. Massive optimal-solution data have been generated for every
possible combination of factors to maximize the responses. |
---|---|
ISSN: | 1220-1758 1841-4303 |