Utilizing decision forest regression machine learning algorithm to predict filling line utilization for optimal maintenance scheduling

Smith, Lauren ; Zastron, Mauritz ; Vlok, P. J. (2018)

CITATION: Smith, L., Zastron, M. & Vlok, P. J. 2018. Utilizing decision forest regression machine learning algorithm to predict filling line utilization for optimal maintenance scheduling. In SAIIE29 Proceedings, 24-26 October 2018, Spier, Stellenbosch, South Africa.

The original publication is available at https://conferences.sun.ac.za/index.php/saiie29/saiie29/schedConf/presentations

Conference Paper

Small margins within the packaging industry mean financial success in this field relies on high equipment availability. To achieve this high equipment availability, maintenance schedules should be carefully planned to minimize downtime. A key component of maintenance schedule planning is predicting equipment utilization. This can prove very difficult as there are many variables such as market demand, seasonality of products, capability and diversity of equipment, and inherent reliability, to name a few. Even some of the leading players in the packaging industry treat the complexities and chaos involved with predicting equipment utilization as a topic best avoided. Current approaches to this problem range from no prediction at all to only a simple linear extrapolation. This paper investigates the merits of using machine learning algorithms to predict equipment utilization in the packaging industry with the aim of optimizing maintenance schedules. Machine learning entails pattern recognition of past data and inclusion of pertinent variables in the present to forecast behaviour. This paper begins with a brief literature review of the field before using data, obtained from a multinational packaging company, to test some of the most promising methods of machine learning in a case study.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/107567
This item appears in the following collections: