基于动态贝叶斯网络的客滚船航行风险预测

Risk Prediction of Ro-ro Passenger Ships Navigation Based on Dynamic Bayesian Network

为识别客滚船航行风险因素,有效预测客滚船航行过程中的航行风险。首先通过历史数据库和相关船舶事故报告识别客滚船航行风险因素,基于识别到的航行风险因素构建故障树模型。然后通过构建的故障树模型确定静态贝叶斯网络结构,引入马尔可夫模型将静态贝叶斯网络结构转化为动态贝叶斯网络结构。最后利用构建的动态贝叶斯网络结构结合琼州海峡徐闻港-新海港客滚航线进行实例分析,计算后验概率预测客滚船航行风险,从而实现客滚船航行风险的预测。结果表明,琼州海峡客滚船航行风险6个时间片的后验概率分别为0.4640、0.5123、0.5457、0.5688、0.5849、0.5961。敏感性分析表明,管理因素对客滚船航行安全的影响程度最大。

To effectively predict navigation risks during the voyage of Ro-Pax(Ro-ro passenger)ships and identify risk factors,this paper first identifies the risk factors of Ro-Pax ships navigation through a historical database and related ship accident reports.A fault tree model is then constructed based on these identified navigation risk factors.The static Bayesian network structure is determined through the constructed fault tree model,and the Markov model is introduced to transform the static Bayesian network structure into a dynamic Bayesian network structure.Finally,an example analysis is conducted using the constructed dynamic Bayesian network structure combined with the Xuwen port to Xinhai port Ro-Pax route in the Qiongzhou Strait.The posterior probability is calculated to predict the Ro-Pax ships navigation risk,achieving the goal of predicting the Ro-Pax ships navigation risk.The results show that the posterior probabilities of six time slices of Ro-Pax ships navigation risk in the Qiongzhou Strait are 0.4640,0.5123,0.5457,0.5688,0.5849,and 0.5961,respectively.Sensitivity analysis indicates that management factors have the greatest impact on the navigation safety of Ro-Pax ships.