Application of machine learning methods on dynamic strength analysis for additive manufactured polypropylene-based composites

This study aimed at applying machine learning (ML) methods to analyze dynamic strength of 3D-printed polypropylene (PP)-based composites. The dynamic strength of additive manufactured PP-based composites with different fillers and printing parameters was investigated by split Hopkinson pressure bars. Based on experimental results, six machine learning approaches were applied to express the relationships between the dynamic strength and materials as well as printing parameters. The performance of the six machine learning algorithms with relatively small training datasets was evaluated. The comparison results showed that artificial neural network could achieve the highest prediction accuracy but with relatively low computational efficiency, whereas the support vector regression could provide satisfactory prediction with both good accuracy and efficiency. The extreme gradient boosting and random forest approaches were recommended if the importance of input was required.

Automated summary

This study applied machine learning methods to analyze the dynamic strength of 3D-printed polypropylene-based composites, evaluating six different algorithms for their performance. Artificial neural network showed high prediction accuracy but low computational efficiency, while support vector regression provided satisfactory prediction with good accuracy and efficiency. Extreme gradient boosting and random forest approaches were recommended for importance of input.