Subway trains operate with high passenger density and complex environments. In collisions, anti-climb energy absorbers may bend under uncertain boundaries, leading to crashworthiness degradation. To clarify this degradation mechanism, an explainable machine learning framework is proposed. First, the sources of uncertain boundary conditions are systematically analyzed, and representative collision parameters are extracted. A quadratic sampling strategy based on local response entropy is developed to construct the collision dataset. Several machine learning models are employed to fit the mapping between boundary conditions and energy-absorption performance, with SHapley Additive exPlanations used to provide explainability of feature contributions. Research findings indicate that absorber alignment and lateral slip are critical factors affecting crashworthiness. Under the most severe boundary conditions, energy absorption decreased by 43.6%. Under single boundary variation, energy absorption exhibits a nonlinear decline as boundary conditions deteriorate. However, under coupled boundary variations, the impact of different boundaries on collision safety shows an intertwined positive and negative pattern. Specifically, when the lateral tilt angle is large, energy absorption first increases and then decreases as lateral displacement increases. This is primarily attributed to the inability of anti-creep teeth to effectively constrain lateral slip. Further explainable analysis quantified the relative contributions of boundary conditions, revealing that vertical displacement is the dominant factor causing collision safety degradation, accounting for over 48% of the contribution. These findings provide theoretical insights and data-driven support for optimizing the design of subway anti-climb energy absorbers.
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号