基于小样本案例的自然灾害风险评估——信息扩散概率模型  

AN UNSYMMETRICAL INFORMATION DIFFUSION PROBABILITY MODEL-FOR RISK EVALUATION OF NATURAL DISASTER BASED ON RARE SAMPLE CASES

作　　者：张韧[1,2] 徐志升[3] 申双和[2] 陈海山[2] 

机构地区：[1]解放军理工大学气象海洋学院军事海洋环境军队重点实验室,南京211101 [2]南京信息工程大学气象灾害省部共建教育部重点实验室,南京210044 [3]成都军区气象中心,成都610051

出　　处：《系统科学与数学》2013年第4期445-456,共12页Journal of Systems Science and Mathematical Sciences

基　　金：气象灾害教育部重点实验室开放课题(KLME1112);国家自然科学基金项目(41075045)资助课题

摘　　要：针对自然灾害(特别是强灾害事件)破坏性大、发生概率小、信息难以获取等制约灾害评估和风险分析的困难,基于信息扩散思想和实际可用小样本案例,提出了一种更具普遍意义的非均匀信息扩散方法,发展了"概率式"信息扩散评估模型.基于该方法对模拟海盗袭击事件的海洋环境风险进行了评估实验.结果表明,该方法可从有限的小样本案例中客观拟合出数据信息的扩散函数,进而实现有限样本数据信息的合理拓展,为自然灾害风险评估中广泛存在的信息不完备问题和困难,探索了一条新的解决方法和技术途径.Aiming at the difficulties of natural disaster （especially strong disaster incidents） violent devastation, less probability and rare collectable information, which restricting risk analysis and disaster evaluation, based on information diffusion idea and actual useable little cases, a sort of more universal unsymmetrical information dif- fusion idea is presented, and an improved ＂probability＂ evaluation model is developed in this paper. The assessment technique is used to evaluate the oceanic environment risk of pirate raid accident, and the experimental result shows that it can objectively fit a suitable diffusion function of data structure from finite sample cases, and carry out the reasonable spread of rare sample cases information by using this ＂probability＂ evaluation model, and a new technique route is introduced for resolving the actual difficults of imperfect information in natural disasters risk evaluation.

关 键 词：小样本案例 信息扩散 概率模型 风险评估 

分 类 号：O242.1[理学—计算数学] X43[理学—数学]