1Faculty of Manufacturing Technologies with a seat in Presov, Presov, Slovak Republic
2Institute of Automation, Robotics and Mechatronics, FME, Technical University of Kosice, Slovak Republic
Journal of Automation and Control.
2015,
Vol. 3 No. 3, 96-100
DOI: 10.12691/automation-3-3-11
Copyright © 2015 Science and Education PublishingCite this paper: Kamil Židek, Vladislav Maxim. Diagnostics of Product Defects by Clustering and Machine Learning Classification Algorithm.
Journal of Automation and Control. 2015; 3(3):96-100. doi: 10.12691/automation-3-3-11.
Correspondence to: Kamil Židek, Faculty of Manufacturing Technologies with a seat in Presov, Presov, Slovak Republic. Email:
kamil.zidek@tuke.skAbstract
The article deals with usability of clustering and machine learning classification algorithm for search systematic surface errors. The main idea is to propose a methodology for the automated identification, diagnostics and localization of systematic errors in mass production. The introduced methodology consists of three levels: image processing for error parameterization, clustering for creating of errors classes (teach data) and prediction of new samples by machine learning algorithm. We conducted experiments with density based clustering algorithm DBSCAN. For classification we use multilayer perceptron MLP/ANN.
Keywords