基于可能性决策的敌对危机行动预测与仿真  

Forecast and Simulation of Adversary Action in Crisis Based on Possibilistic Decision Method

作　　者：胡睿[1] 柳少军[1] 

机构地区：[1]国防大学信息作战与指挥训练教研部,北京100091

出　　处：《指挥控制与仿真》2013年第5期70-74,共5页Command Control & Simulation

基　　金：国家自然科学基金(61203140)

摘　　要：可能性决策能较好地应对军事领域常常出现的概率难以获得,或者强调"出奇制胜"而有意规避概率风险的情形,是解决知识不完备情况下敌对行动预测的有力手段。从可能性理论基本公理体系出发,结合动态规划方法,提出并证明可能性决策的最优化定理,给出多步条件下可能性决策的实现算法,结合危机条件下敌对行动预测的范例和仿真分析,并与传统概率风险决策进行比较,体现可能性决策的优越性,为该方法的推广应用创造条件。The possibilistic method is expected to be a powerful means to solve the multi-stage decision problem under the conditions that the knowledge is far from complete,for it is preferable to probabilistic method in the case that probability is hard to know or the adversary intentionally ignores probability-based risk to beyond expectation.Starting from the basic definition of the theory of probability and the axiom system and combined with the principle of dynamic programming,an optimization theorem of possibilistic decision is proposed and proved,as a result,the realization algorithm of multi-step decision is brought out.Finally,the possibilistic decision method is applied to an example of adversay action forecast in crisis with simulation to show how the method works and where is the superiority.

关 键 词：可能性决策 敌对行动预测 最优性定理 动态规划 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] E917[自动化与计算机技术—计算机科学与技术]