Machine Learning Algorithms for Predicting Mechanical Stiffness of Lattice Structure-Based Polymer Foam

Area of Science:

• Materials Science
• Mechanical Engineering
• Computational Science

Background:

• Polymer foams offer good mechanical and energy absorption but are hard to produce consistently.
• Lattice structures, manufacturable via additive manufacturing, allow for greater design control over material properties.

Purpose of the Study:

• To investigate the influence of lattice parameters on the mechanical stiffness of additively manufactured lattice parts.
• To compare the effectiveness of various machine learning algorithms in predicting lattice part stiffness.

Main Methods:

• Designed 360 lattice parts varying lattice type, cell dimensions (X, Y, Z), and cell wall thickness.
• Performed computational analyses under consistent loading conditions to record strain.
• Compared Linear Regression, Polynomial Regression, Decision Tree, Random Forest, and Artificial Neural Network (ANN) for prediction accuracy.

Main Results:

• All machine learning algorithms showed low prediction errors (MSE, RMSE, MAE).
• Artificial Neural Network (ANN) demonstrated superior performance with a correlation coefficient of 0.93.
• ANN's accuracy was validated through relative error analysis and actual vs. predicted value plots.

Conclusions:

• Machine learning, particularly ANN, can accurately predict the mechanical stiffness of lattice parts based on design parameters.
• This predictive capability facilitates the optimized design of lattice structures for desired mechanical properties.