Bionic horsetail-honeycomb tube: design, crashworthiness analysis and data-driven optimization

Thin-walled tubular energy absorbers require balance between load-bearing capacity, energy absorption efficiency and lightweighting. This study proposes a bionic horsetail-honeycomb thin-walled tube (BHHT) merging the multi-chamber configuration of horsetail stems with the energy-absorbing topology of hexagonal honeycombs. Specimens were fabricated via selective laser melting and tested under quasi-static compression. The validated finite element model achieved simulation errors below 8% for EA, SEA, and IPCF. Coupling analysis revealed that the overall EA of BHHT reached 142% of the algebraic sum of its substructures. Parametric studies identified complex nonlinear interactions among geometric parameters. A three-stage machine learning optimization framework was developed: a dataset was constructed using Latin hypercube sampling and finite element simulations; a high-precision surrogate model (R 2 > 0.97, MAPE < 4.5%) was trained and interpreted to quantify parameter contributions; NSGA-II with combined-weighting TOPSIS was then employed to maximize EA and SEA while minimizing IPCF. The optimal design (n = 10, t = 0.61 mm, d = 26.52 mm, l = 2.51 mm) yielded EA = 4123 J, IPCF = 108.67 kN, and SEA = 51.81 kJ·kg⁻ 1 . Compared with the original design, IPCF decreased by 32.84% and SEA increased by 8.32%, achieving synergistic optimization of loading smoothness, energy absorption efficiency, and lightweight performance. This study provides a methodology for the design and data-driven optimization of bionic energy-absorbing structures.