摘要
为解决现有的山区暴雨-地质、市区暴雨-洪涝灾害模型在链式反应推理和精度上的不足,基于贝叶斯网络理论分析暴雨灾害演化规律,构建灾害链推理模型;根据国内暴雨灾害历史数据和相关文献总结选取暴雨灾害节点变量,构建暴雨灾害链拓扑结构,并应用期望最大化算法求得暴雨灾害条件概率,基于因果推理实现对暴雨引发的次生灾害和基础设施损毁等级的预测。最后以湖南省宁远县2017年6月22日至7月1日间因暴雨导致的滑坡洪灾为例,运用上述暴雨-地质、暴雨-洪涝灾害链推理模型进行实例验证,预测结果与实际情况吻合较好,Brier检验的B值小于0.6,结果表明该灾害链推理模型具有可行性。
In order to solve the problems of chain reaction reasoning and precision in the existing models of rainstormgeology and urban rainstorm-flood disaster in mountainous areas,this paper analyzes the evolution law of rainstorm disaster based on Bayesian network theory and constructs the reasoning model of disaster chain.Based on the historical data of rainstorm disaster and related literature,the paper selects the node variables of rainstorm disaster,constructs the topology of rainstorm disaster chain,and uses expectation maximization algorithm to obtain the probability of rainstorm disaster condition,and based on causality reasoning,the prediction of secondary disasters and infrastructure damage caused by rainstorm is realized.Finally,the paper takes the landslide flood caused by heavy rain from June 22 to July 1,2017 in Ningyuan County,Hunan province as an example,the prediction results are in good agreement with the actual situation,and the B value of the Brier test is less than 0.6.The results show that the disaster chain reasoning model is feasible.
作者
帅敏
郭海湘
刘晓
王德运
陈卫明
Shuai Min;Guo Haixiang;Liu Xiao;Wang Deyun;Chen Weiming(School of Eeonomics and Management,China University of Geosciences(Wuhan),Wuhan 430074,China;School of Engineering,China University of Geosciences(Wuhan),Wuhan 430074,China)
出处
《科技管理研究》
CSSCI
北大核心
2021年第4期191-197,共7页
Science and Technology Management Research
基金
国家自然科学基金项目“基于数据驱动的滑坡地质灾害预测及其应急决策研究——以长江经济带三峡库区为例”(71874165),“大数据驱动下自然资源生态安全预测预警预案研究”(72074198)
教育部哲学社会科学研究后期资助项目“应急救援队伍优化调配与合作救援仿真研究”(20JHQ094)。
关键词
暴雨-洪涝
暴雨-地质
灾害链
贝叶斯网络
rainstorm-flood
rainstorm-geology
disaster chain
Bayesian network
