Activity Number:
|
78
- Contributed Poster Presentations: Caucus for Women in Statistics
|
Type:
|
Contributed
|
Date/Time:
|
Monday, August 3, 2020 : 10:00 AM to 2:00 PM
|
Sponsor:
|
Caucus for Women in Statistics
|
Abstract #312480
|
|
Title:
|
Recurrent Event Data-Analysis for Replacement Parts
|
Author(s):
|
Flavia Dalia Frumosu* and Georg Ørnskov Rønsch and Murat Kulahci
|
Companies:
|
Technical University of Denmark and Technical University of Denmark and Technical University of Denmark;
|
Keywords:
|
Industry 4.0;
Manufacturing;
Predictive maintenance;
Survival Analysis;
Recurrent event data
|
Abstract:
|
As a direct result of Industry 4.0, different industries are facing various challenges in automation, in handling large amounts of data (Big Data) and digitalization. The insights gained from the generated data can help industries towards better understanding of their processes and lead to significant improvement opportunities. The current work comes from a collaboration with an industrial partner dealing with the manufacturing of plastic products using an injection molding process. In the daily work of the industrial partner, different parts of the molding machine are replaced in order to make sure that plastic products are produced within established tolerances. The industrial partner is interested in predicting when new parts need to be replaced given the available historical data. This work is exploring non-parametric recurrent event data-analysis along with machine learning methods to tackle the prediction problem for such an industrial setting. One of the main challenges behind this problem is the combination of existing historical data in an efficient manner for prediction purposes.
|
Authors who are presenting talks have a * after their name.