Risk Factor Identification and Analysis of Grounding Incidents for Bulk Carriers on the Beijiang River Based on Accident Reports

With the increasing volume of inland waterway traffic, grounding incidents involving bulk carriers on the Beijiang River have become more frequent. This study applies natural language processing techniques and causal network modeling to identify and analyze key grounding risk factors based on 54 accident reports from 2019 to 2023. Using a structured text mining pipeline—including TF-IDF weighting, semantic normalization, and chi-square feature selection—40 representative risk factors were extracted. A directed, weighted risk network was constructed, and centrality measures, including improved K-Shell decomposition, degree centrality, and PageRank, were used to evaluate the positional and structural importance of each factor. The results highlight that underestimation of helm response (No. 17), overloaded cargo (No. 18), misjudgment of environmental effects (No. 19), and deviation from planned route (No. 37) are the most influential contributors to grounding incidents. This analysis provides a data-driven foundation for targeted risk management and inland waterway safety improvement.