Tool wear classification using decision trees in stone drilling applications: A preliminary study

Klaić, Miho and Staroveški, Tomislav and Udiljak, Toma (2014) Tool wear classification using decision trees in stone drilling applications: A preliminary study. = Tool wear classification using decision trees in stone drilling applications: A preliminary study. In: 2013 24th DAAAM International Symposium on Intelligent Manufacturing and Automation, 23-26.10.2013., Zadar; Croatia.

[img]
Preview
Text
Dummy article.pdf - Draft Version Jezik dokumenta:English

Download (79kB) | Preview
Official URL: http://www.scopus.com/inward/record.url?eid=2-s2.0...

Abstract

Process parameters of stone drilling with a small diameter twist drill were used to predict tool wear by means of a machine learning decision tree algorithm. The model links tool wear with features extracted from the force sensor and the main and feed drive current sensors signals recorded under different cutting conditions and different tool wear states. Signal features extracted from both the time and frequency domain were used as input parameters for construction of a decision tree which classifies the tool state into sharp or worn. The model was refined by selecting only the feature sources most important for classification. The best model achieves 90 accuracy in classification and relies only on features of the current signals, which simplifies its implementation in a CNC system for industrial applications. © 2014 The Authors. Published by Elsevier Ltd.

Item Type: Conference or Workshop Item (Lecture)
Keywords (Croatian): Machine learning; Stone drilling; Tool condition monitoring; Tool wear
Subjects: TECHNICAL SCIENCE > Mechanical Engineering
Divisions: 1200 Department of Technology > 1230 Chair of Machine Tools
Indexed in Web of Science: No
Indexed in Current Contents: No
Citations SCOPUS: 1 (21.5.2015.)
Date Deposited: 21 May 2015 09:24
Last Modified: 21 May 2015 09:24
URI: http://repozitorij.fsb.hr/id/eprint/4311

Actions (login required)

View Item View Item

Downloads

Downloads per month over past year

Nema podataka za dohvacanje citata