基于贝叶斯网络的海盗袭击事件影响因素

Influencing Factors of Piracy Attacks with Bayesian Network

为分析海盗袭击事件的影响因素,减少海盗袭击事件的发生,保障海上运输安全,引入贝叶斯网络理论,利用期望最大化(Expectation Maximization,EM)算法和联合树推理算法构建海盗袭击事件分析模型。根据东南亚海域海盗袭击事件的实际数据获得贝叶斯网络(Bayesian Networks, BN)结构各节点之间的概率依赖关系,并对海盗袭击成功节点进行敏感性分析。分析结果表明:外界环境、船舶自身风险和时间是影响海盗袭击事件发生的主要因素。

The Bayesian network theory is introduced to analyze the influencing factors of the pirate attacks. The analytic model of piracy attacks is established. The EM(Expectation Maximization) algorithm and the joint tree algorithm are used for determining the structure parameters and conducting sensitivity analysis. Case study for Southeast Asian sea area shows that those nodes of external environment, ship’s own risk and time are of the most significances with regard to the influences on the probability nodes of piracy attacks.