基于贝叶斯网络的船舶溢油风险评价研究

Risk Assessment of Ship Oil Spill Based on Bayesian Network

船舶溢油事故已成为导致海洋污染重要的因素之一,采用科学方法对船舶溢油风险进行有效的预测与评估具有重要意义。将船舶溢油风险分为操作性溢油风险与事故性溢油风险两类,通过分析历史数据与借助专家经验识别风险因素,构建了船舶溢油风险的贝叶斯网络模型和条件概率表CPT,并利用HUGIN软件进行了概率推理和风险因素灵敏度分析,定量评估了船舶溢油风险,找出了影响最突出的风险因素。将贝叶斯网络模型应用于我国沿海港口水域,得出两类船舶溢油风险概率分别为0.013 8和0.000 3,指出了加燃油、装卸油品、人员疏忽和船舶密度等风险因素对船舶溢油风险影响最突出。

The accidents of ship oil spill have been one of the most significant factors causing marine pollution,so appropriate approaches to forecast ship oil spill risk has important significance.A novel Bayesian model for risk assessment of ship oil spill is presented.Ship oil spill is divided into two kinds,i.e.＇operation spill＇ and ＇accident spill＇.The Bayesian network model and the statistics CPT are constructed by analyzing historic accident data and expert experience.The software HUGIN is used to compute accident probability of ship oil spill and to analyze sensitivity of risk factors so as to evaluate ship oil spill risks quantitatively and find out main risk factors.The presented Bayesian network model is applied to study the ship oil spill risk in China＇s coastal harbor waters,which led to the conclusion that probabilities for the two ship oil spill kinds are 0.013 8 and 0.000 3 respectively and that bunkering,loading or discharging operation,personnel negligence and ship density etc.are the major factors which influence ship oil spill risks.