Machine learning-enabled optimization of extrusion-based 3D printing.

Area of Science:

• * Combines rapidly advancing fields of machine learning (ML) and three-dimensional (3D) printing.
• * Focuses on optimizing manufacturing processes through computational intelligence.

Background:

• * 3D printing enables complex structure manufacturing but parameter optimization remains challenging.
• * Current methods lead to increased pre-printing time and material waste.
• * Existing research often uses orthogonal design, limiting comprehensive parameter exploration.

Purpose of the Study:

• * To develop the first graphical user interface (GUI) integrating ML and 3D printing for parameter optimization.
• * To introduce a novel method for calculating a
• complexity index
• for computer-aided design (CAD) models.
• * To identify the most effective ML model for predicting 3D print outcomes.

Main Methods:

• * Utilized nine diverse CAD images and developed a custom method to quantify design complexity.
• * Employed four novel labeling methods (manual and automatic) to assess print outcome similarity to CAD designs.
• * Trained eight ML algorithms on 224 data points, evaluating their predictive performance.

Main Results:

• * The gradient boosting regression model achieved the highest prediction accuracy with an R-2 score of 0.954.
• * The developed ML-embedded GUI allows users to input designs and printing parameters (temperature, pressure).
• * The system provides an approximate similarity prediction for the intended 3D print.

Conclusions:

• * The integrated ML-GUI system significantly reduces trial-and-error in 3D printing parameter selection.
• * This approach accelerates the design-to-product timeline.
• * It leads to substantial reductions in material waste and enhances cost-efficiency for both skilled and unskilled users.