基于贝叶斯网络的社会安全风险演化分析

Evolutionary analysis of social security risks based on Bayesian networks

为提高社会安全风险防控能力,本文基于贝叶斯网络构建一种适用于社会安全风险演化分析的事件链模型。首先,将社会安全风险演化过程中的突发事件作为基本单元——元事件;其次,以贝叶斯网络为建模工具,关联合并元事件贝叶斯网络;最后,选取爆炸、火灾、坍塌、泄漏、暴雨和群体性事件6种不同元事件,构建贝叶斯网络模型。以辽宁省大连市2011年8月14日大型群众游行事件为例模型验证,结果表明:基于贝叶斯网络构建的社会安全风险演化事件链模型具有可行性,可根据贝叶斯正向因果推理预测不同证据组合下社会安全风险的发生概率,推断社会安全风险的发展演化态势。

In order to improve the ability of social security risk prevention and control,an event chain model applicable to the analysis of social security risk evolution is constructed based on Bayesian network.Firstly,the emergency in the process of social security risk evolution are all regarded as basic units-meta-events.Then,Bayesian network is used as a modelling tool to associate and merge the meta-event Bayesian network.Finally,six different meta-events,namely,explosions,fires,collapses,leaks,rainstorms,and mass events,are selected to construct the Bayesian network model,and the model is validated with the Dalian City,Liaoning Province,14 August 2011 large-scale mass procession event as an example for model validation.The results show that the event chain model based on Bayesian network has feasibility in practical application,and it can predict the probability of occurrence of social security risks under different combinations of evidence according to Bayesian forward causal inference,and infer the development and evolution of social security risks.