Abstract
Rapid changes in technology have led to an increasingly fast pace of product introductions. For long-life systems (e.g., planes, ships, and nuclear power plants), rapid changes help sustain useful life, but at the same time, present significant challenges associated with obsolescence management. Over the years, many approaches for forecasting obsolescence risk and product life cycle have been developed. However, gathering inputs required for forecasting is often subjective and laborious, causing inconsistencies in predictions. To address these issues, the objective of this research is to develop a machine learning-based methodology capable of forecasting obsolescence risk and product life cycle accurately while minimizing maintenance and upkeep of the forecasting system. Specifically, this new methodology enables prediction of both the obsolescence risk level and the date when a part becomes obsolete. A case study of the cell phone market is presented to demonstrate the effectiveness and efficiency of the new approach. Results have shown that machine learning algorithms (i.e., random forest, artificial neural networks, and support vector machines) can classify parts as active or obsolete with over 98% accuracy and predict obsolescence dates within a few months.
Original language | English (US) |
---|---|
Article number | 7543522 |
Pages (from-to) | 1428-1439 |
Number of pages | 12 |
Journal | IEEE Transactions on Components, Packaging and Manufacturing Technology |
Volume | 6 |
Issue number | 9 |
DOIs | |
State | Published - Sep 2016 |
All Science Journal Classification (ASJC) codes
- Electronic, Optical and Magnetic Materials
- Industrial and Manufacturing Engineering
- Electrical and Electronic Engineering