基于数据驱动贝叶斯网络的内河船舶交通事故分析

Traffic Accident Analysis of Inland Waterway Vessels Based on Data-driven Bayesian Network

伴随着我国内河航运的蓬勃发展,内河船舶的行驶安全受到了广泛的关注。将贝叶斯网络引入内河船舶交通事故分析,探究不同事故类型与航道、气候、船舶等方面影响因素之间的关系。首先,以芜湖海事局2013—2020年间上报的219条内河航运事故船舶数据为基础,抽取样本并使用贝叶斯可视化软件Netica训练得出内河船舶交通事故的贝叶斯网络模型;其次,运用贝叶斯推理提炼出不同事故类型发生的关键影响因素,并对模型训练结果的准确性进行验证;最后,将情景分析法引入到模型中,预测不同影响因素组合波动时最有可能发生的内河船舶交通事故类型,可为海事管理部门采取有针对性的防控措施提供理论支撑。

In the context of the vigorous development of inland waterway shipping in China, people pay more attention to traffic safety of inland waterway vessels.This paper introduces the Bayesian Network into the analysis of inland waterway vessel traffic accidents to explore the relationship between different accident types and influencing factors such as waterway, climate, and ships.Firstly, with 219 inland waterway vessel accident data reported by Wuhu Maritime Safety Administration from 2013 to 2020 as the database, samples are selected and trained with the Netica software to establish the Bayesian Network model of inland waterway vessels.Secondly, the key factors of different accident types are extracted through Bayesian inference, and the accuracy of model training result is verified.Finally, the most possible types of inland waterway vessel accidents with the combination of different influencing factor fluctuations are predicted by the scenario analysis method. The research can provide theoretical support for the maritime management departments to take targeted measures.