基于贝叶斯网络的城市地震次生灾害演化机理分析

摘要

In order to understand the evolution routes and diffusion mechanism of earthquake-induced urban disasters and provide reference for seismic surveillance and prediction, node variables and their value ranges of the Bayesian network for evolution system of earthquake-induced urban disasters are determined based on analysis on relative data and literatures, and a Bayesian network structure diagram is established according to the causality of node variables. Conditional probability of each variable in the network is acquired by data statistics of 16 cities affected by representative earthquakes at home and abroad. Finally, posterity probabilities of secondary disasters under the condition of the same input variables and different environment variables are obtained by using Bayesian network toolbox based on Bayesian network reasoning. The results of an illustrated example show that the level of emergency management has an important impact on the occurrence probability of different earthquake-induced urban disasters.%为了解城市地震次生灾害的演化路径及扩散机理,进而为城市防震救灾工作提供科学的参考依据,通过对城市地震有关资料和文献的分析,确定出城市地震次生灾害演化系统贝叶斯网络的节点变量及取值范围,并根据节点变量之间的因果关系构建出贝叶斯网络结构图.通过对国内外典型震例中的16个受灾城市相关数据进行统计,分析获得网络结构各个变量的条件概率.结合贝叶斯网络推理知识,利用贝叶斯网络工具箱分析了在相同的地震输入变量、不同的控制输入变量取值条件下,各类城市地震次生灾害的发生概率.算例分析结果表明不同的应急管理水平对城市地震各类次生灾害的发生概率有显著影响.