Integrated multi-objective optimization on the geometrical design of a disk-type milling cutter with multiple inserts applying uniform design, RBF neural network, and PSO algorithm

Achmad Arifin, Yu Ren Wu

Research output: Contribution to journalArticlepeer-review

Abstract

A cutter with inserts is suitably applied in screw rotor milling. However, the cutter design process, including arranging the inserts onto the cutter body accurately, involves numerous factors, so it is costly and lengthy to achieve a precise cutter design if performed by trial error. This study introduces an integrated optimization to simplify the cutter designing process by minimizing multi-objective factors (number of inserts, grinding stock amount, and rotor profile deviation), respecting four critical design factors (grinding allowance, insert arrangement area, insert inclination angle, and correctional offset). The uniform design was applied to obtain uniformity and representativeness in the experiment sample size, whereas radial basis function (RBF) approximated the design factors, and particle swarm optimization (PSO) predicted the optimum results. The result confirms that all objective factors were diminished significantly where the grinding allowance is the most influential factor. In addition, the rotor surface topography indicated a consistent deviation. Finally, the optimized cutter is reliable, and the integrated optimization model is effective and entirely practicable.

Original languageEnglish
Pages (from-to)4829-4846
Number of pages18
JournalInternational Journal of Advanced Manufacturing Technology
Volume121
Issue number7-8
DOIs
StatePublished - Aug 2022

Keywords

  • Milling cutter design
  • Multi-objective optimization
  • Particle swarm optimization
  • Radial basis function
  • Screw rotor milling
  • Uniform design model

Fingerprint

Dive into the research topics of 'Integrated multi-objective optimization on the geometrical design of a disk-type milling cutter with multiple inserts applying uniform design, RBF neural network, and PSO algorithm'. Together they form a unique fingerprint.

Cite this