基于贝叶斯网络的港口干散货货运风险预测研究

Risk Prediction of the Freight Distribution in Dry Bulk Port Using Bayesian Network

建立数据驱动的贝叶斯网络风险预测模型,对港口干散货货运领域风险问题进行研究。实证结果表明:该方法能克服传统风险预测方法对专家领域知识的过度依赖;能有效预测港口干散货货运中的不确定性风险事件;能图形化表示港口干散货货运风险与多种因素间依赖关系,能在众多因素中清晰表示各因素间依赖关系强弱,进而提高港口管理层对港口干散货货运风险的认识,辅助其制定管理决策。为加强货运风险的控制,港方应完善信息采集功能,丰富特征集,提升货运风险预测能力;港方和港口公安局可根据实证结果优化车辆抽查策略,有的放矢。

A data-driven Bayesian network risk prediction modeling method is proposed for analyzing the risk of freight distribution in dry bulk port.Empirical results indicate that,the data-driven modeling method proposed can address the limitation of the immoderate reliance on experts'judgements in traditional methods of risk analysis.It can also predict the risk events of uncertainty effectively,represent the dependencies between the risk and various factors graphically and express the dependency degree clearly.Furthermore,it can be used to improve managers'understanding of the risk of freight distribution in dry bulk port and provide decision-making support.In order to strengthen the control on freight distribution risks,port management side should improve the information collection function of the logistic system,enrich the feature sets and enhance risk prediction ability;together with port public security bureau,it can optimize vehicle inspection strategies based on empirical results.