Machine learning driven optimization of compressive strength of 3D printed bio polymer composite material

Area of Science:

• Materials Science
• Additive Manufacturing
• Machine Learning

Background:

• 3D printing enables complex multi-layered designs.
• Functionally Graded Materials (FGMs) offer tailored properties.
• Optimizing FGM mechanical strength is crucial for applications.

Purpose of the Study:

• To optimize the compressive strength (CS) of PLA/Almond Shell Reinforced PLA FGMs.
• To evaluate machine learning (ML) models for predicting CS based on FFF process parameters.
• To compare ML optimization with the traditional Taguchi method.

Main Methods:

• Fabrication of FGMs using the Fused Filament Fabrication (FFF) process.
• Utilized six ML models to predict CS based on print speed (PS), layer height (LH), and printing temperature (PT).
• Employed SHAP analysis to determine parameter influence and Polynomial Regression (PR) for optimization.

Main Results:

• Polynomial Regression (PR) achieved the best prediction accuracy (R²=0.88) among ML models.
• Print speed (PS) and layer height (LH) were identified as the most influential parameters.
• Optimized parameters predicted a CS of 36 MPa, experimentally validated at 34.8 MPa (3.44% error).

Conclusions:

• PR-based ML optimization significantly improves accuracy and mechanical performance of FGMs.
• The optimized FGMs demonstrate suitability for various consumer applications.
• ML optimization surpasses traditional methods like Taguchi, offering enhanced predictive power and material performance.