Using artificial neural networks to forecast operation times in metal industry


Kumru M., Kumru P.

INTERNATIONAL JOURNAL OF COMPUTER INTEGRATED MANUFACTURING, vol.27, pp.48-59, 2014 (Journal Indexed in SCI) identifier identifier

  • Publication Type: Article / Article
  • Volume: 27 Issue: 1
  • Publication Date: 2014
  • Doi Number: 10.1080/0951192x.2013.800231
  • Title of Journal : INTERNATIONAL JOURNAL OF COMPUTER INTEGRATED MANUFACTURING
  • Page Numbers: pp.48-59

Abstract

This study was conducted in an auto spare parts production plant where the biggest bottleneck in managing the enterprise is the lack of fulfilling the requisitions of the customers on time. The main reason for the delay is the absence of operation time data valid for the required parts ordered with different specifications. For preparing effective schedules and for eliminating the bottleneck, the factory needs to use reliable operation time data for each part produced. An artificial neural network (ANN) approach was used for this purpose, and its forecasting performance was compared with that of multiple linear and nonlinear regression models. Based on the statistical analyses, the ANN approach outperformed the regression models and is found to be more reliable in forecasting the operation times.