Work

Machine learning‐based time series analysis of polylactic acid bead foam extrusion

Year: 2024

Type: article

Abstract: Understanding the behavior of polymer melts during extrusion is essential for optimizing processes and developing new materials. However, analyzing the continuous data generated by an extruder poses significant challenges. This paper investigates the utility of machine learning in predicting melt pressure at the die plate in polylactic acid (PLA) bead foam extrusion, a critical parameter in the extrusion process. Utilizing a random forest (RF) model, we examine how various processing parameters influence melt pressure. By segmenting the data into time‐delayed intervals, we achieve accurate predictions. We present forecasts of melt pressure at the die for intervals of 5 s, 1 min, and 5 min, demonstrating particularly strong performance for the 5‐min forecast with a Mean Absolute Error (MAE) of 1.88 and the coefficient of determination ( score) of 0.90. By exploring time series data, our study demonstrates the effectiveness of the RF model and provides a foundation for more advanced and precise control strategies in polymer bead extrusion processes.

Source: Journal of Applied Polymer Science

DOI: 10.1002/app.56170

Authors: Karim Ali Shah, Rodrigo Q. Albuquerque, Christian Brütting, Holger Ruckdäschel

Institutions: University of Bayreuth, Neue Materialien Bayreuth GmbH

Language: English

Topic: Machine Learning in Materials Science

Subfield: Materials Chemistry

Field: Materials Science

Domain: Physical Sciences

Open Access status: hybrid

Estimated APC paid: $4,030

Funders: Deutsche Forschungsgemeinschaft, Siemens, Bayerisches Staatsministerium für Wissenschaft und Kunst