Crashworthiness evaluation and optimization of full polypropylene sandwich tubes under low-velocity impact based on machine learning algorithms
Structures, vol.60, 2024 (SCI-Expanded, Scopus)
- Publication Type: Article / Article
- Volume: 60
- Publication Date: 2024
- Doi Number: 10.1016/j.istruc.2024.105901
- Journal Name: Structures
- Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
- Keywords: Bio-inspired Sandwich Tube, Energy Absorption, Machine Learning, MLP, Optimization, Polypropylene
- Çukurova University Affiliated: Yes
Abstract
The performance of traditional single and multi-cell tubular energy absorbers was enhanced through the design of a polypropylene tube inspired by biological structures such as horsetail and tendons. In this study, the energy-absorbing behavior of these tubes was investigated, and the influence of parameters like thickness and height was examined. Using the LS-DYNA finite element code, various specimens with different thicknesses, heights, and core counts were modeled. Crashworthiness predictions for these designs were made using machine learning, specifically the Multi-layer Perceptron (MLP) algorithm. The optimal sandwich tube, in terms of Performance Criteria Factor (PCF) and Specific Energy Absorption (SEA), was identified using the non-dominated sorting genetic algorithm II (NSGA-II). The specimen featuring three core tubes, 1.1 mm thickness, and 82 mm height was determined to have the most efficient performance. Its force-displacement curve was produced using MLP methods and the Bayesian Regularization training algorithm. Lastly, the crashworthiness behavior of this optimized bio-inspired sandwich tube was validated through experimental testing and finite element modeling.