Decision Tree Methods for Predicting Surface Roughness in Fused Deposition Modeling Parts

Area of Science:

• Additive Manufacturing
• Materials Science
• Data Science

Background:

• Fused Deposition Modeling (FDM) involves numerous control parameters influencing part quality.
• Predicting the final surface finish of FDM-printed parts based on these parameters is complex.
• Surface roughness is a critical factor in the functional and aesthetic performance of 3D printed components.

Purpose of the Study:

• To compare the predictive performance of different decision tree algorithms for surface roughness in FDM-printed parts.
• To identify the most effective decision tree model for predicting surface roughness in polyethylene terephthalate glycol (PETG) parts.
• To analyze the relationship between FDM control parameters and the resulting surface roughness of PETG components.

Main Methods:

• Development and comparison of decision tree models including C4.5, random forest, and random tree algorithms.
• Utilizing a dataset of 27 instances with FDM parameters: layer height, extrusion temperature, print speed, print acceleration, and flow rate.
• Evaluation of model performance using an independent dataset of 15 instances.

Main Results:

• The random tree algorithm demonstrated superior performance in predicting surface roughness.
• All decision tree models showed varying degrees of success in forecasting surface finish.
• The study successfully correlated specific FDM parameters with surface roughness outcomes.

Conclusions:

• The random tree algorithm is the most effective for predicting surface roughness in FDM-printed PETG parts.
• Decision tree models offer a viable approach to optimize FDM process parameters for desired surface finish.
• Further research can explore more complex models and a wider range of materials and parameters.