Ensuring Part Quality for Material Extrusion by Developing a Methodology for Use-Case-Specific Parameter Set Determination Using Machine Learning Models

Abstract

The material extrusion of plastics has matured into a lucrative and flexible alternative to conventional manufacturing. A major downside of this process is the missing quality assurance caused by the influence of process parameters on part quality. Such parameters—e.g., infill density and print speed—are selected prior to manufacturing. As a result, the achieved part quality is mostly unknown, limiting the use of material extrusion and leading to increased material costs and print times. A promising approach to overcome this drawback are prediction models, especially methods of machine learning. Yet, a methodology that enables their integration in the manufacturing process is lacking. This paper provides a methodology based on a lookup approach and calculated safety factors. The methodology is tested and subsequently applied to two exemplary use cases. The result empowers users and researchers with a methodology to use prediction models for quality assurance in their company environment. On the other hand, future improvements and new research results can be integrated into the methodology to verify its applicability in practice.