Parameter optimization for NC machine tool based on golden section search driven PSO

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

13 Scopus citations

Abstract

We have proposed a modified PSO[1]; GPSO (Golden-section-search driven Particle Swarm Optimization) which updates only one particle in a generation based on a strategy: golden section search and steepest descent method. It was proved to be effect in various optimization problem. In this paper, first, this GPSO is revised to make clear its effectiveness. Then, the GPSO is utilized to optimize control parameters in NC machine tools. Parameters which are said to be difficult to optimize in a NC machine tool, is chosen and the roles of those parameters are scrutinized. Based on those scrutiny, fitness are defined for parameters. In order to verify optimization performance of the algorithms (GA, PSO, GPSO), a hardware-in-the-loop system with a NC machine tool is set up and on-line optimization experiments are conducted using the system. In experiments, the GPSO shows better optimization performance.

Original languageEnglish
Title of host publication2007 IEEE International Symposium on Industrial Electronics, ISIE 2007, Proceedings
Pages3114-3119
Number of pages6
DOIs
StatePublished - 2007
Event2007 IEEE International Symposium on Industrial Electronics, ISIE 2007 - Caixanova - Vigo, Spain
Duration: 4 Jun 20077 Jun 2007

Publication series

NameIEEE International Symposium on Industrial Electronics

Conference

Conference2007 IEEE International Symposium on Industrial Electronics, ISIE 2007
Country/TerritorySpain
CityCaixanova - Vigo
Period4/06/077/06/07

Keywords

  • Golden section search
  • Golden section search driven particle swarm optimization (GPSO)
  • Hardware-in-the-loop system
  • Parameter tuning
  • Precision motion control
  • Steepest descent method

Fingerprint

Dive into the research topics of 'Parameter optimization for NC machine tool based on golden section search driven PSO'. Together they form a unique fingerprint.

Cite this