Abstract: Ship collisions pose significant risks to ship structures and the safety of lives on board. Rapid analysis of the extent of structural damage from collisions can provide a critical basis for emergency risk mitigation and damage control and rescue operations. While the Finite Element Method（FEM） can accurately calculate the degree of structural collision damage, it is time-consuming and requires numerous parameters, making it unsuitable for rapid damage assessment under limited input conditions. Using a dataset of 202 ship collision accidents that occurred between 2015 and 2019, this study applies statistical analysis to establish relationships between damage factors and damage levels. A Bayesian network model is developed to analyze the risk of collision damage grades under the combined influence of multiple factors. The results demonstrate that the proposed method agrees well with actual accident cases and can rapidly estimate the collision damage level of ship structures even when only limited parameters are available.