基于贝叶斯网络的突发事件链建模方法与应用

摘要

According to the characteristics of emergent events which are highly uncertain, systematization and cascading, the paper abstracted emergent events as a three-layer structure model, including input, state and output, based on system theory. A single event Bayesian network for uncertain emergent event was put forward. Then, a combination method for a succession of associated Bayesian networks of single event was proposed, on the basis of which, a Bayesian network of emergent event chain was built. Finally, support for feasibility and validity of the model proposed was provided by applying it to an example of a typhoon-rainstorm-flood emergent event chain.%针对各突发事件具有高度不确定性、系统性和链式反应的特点,将各类突发事件统一抽象为输入、状态和输出相互作用的三层网络结构.以贝叶斯网络为建模工具,提出了适用于不确定性突发事件的单一事件贝叶斯网络.在此基础上,提出了具有关联关系的单一事件贝叶斯网络的合并方法,进而构建出了突发事件链贝叶斯网络模型.最后,以台风-暴雨-洪水事件链为例验证了该方法的可行性和有效性.