Latent-driven structural generation and performance-constrained optimization for expansion tube energy absorbers

Abstract

The expansion energy absorption structures are widely used in engineering fields such as rail transportation due to its controllable deformation and strong energy dissipation capacity. However, complex structural parameters and boundary conditions will cause various nonlinear deformation modes, affecting structural performance evaluation and design optimization. Therefore, this paper proposes a unified modeling framework to achieve integrated analysis of structure generation and performance prediction. The framework generates high-fidelity deformed images based on design parameters, and identifies the deformation patterns and physical consistency of the images through a classification network. Meanwhile, a perception regression model is constructed to predict key crashworthiness indicators and achieve rapid estimation of structural performance. On this basis, high-dimensional sampling and performance-based structural selection methods are combined to extract the best structural solution in each deformation mode. The research results verify the effectiveness of the framework in deformation mode prediction and structural design, and provide important guidance for the optimization of energy-absorbing structures.