任务规划中基于案例推理的高维解空间适应性问题研究  

The Research of High Dimensional Solution Space Adaptation Based on Case-Based Reasoning during Mission Planning

作　　者：张媛[1] 齐玉东[1] 乔勇军[1] 陈青华[1] 

机构地区：[1]海军航空工程学院兵器科学与技术系,山东烟台

出　　处：《计算机科学与应用》2015年第12期454-463,共10页Computer Science and Application

基　　金：国家自然科学基金。

摘　　要：利用案例推理对指挥实体任务规划过程中决策问题求解方法的修正过程是该方法推理过程中最困难的阶段,尤其当决策问题解空间是多维的情况下。文章讨论了指挥实体任务规划过程中高维决策空间的修正问题,并提出了可行的解决方法。首先利用自组织匹配法(ViSOM)清晰展现问题空间与决策空间的映射过程,然后,利用BP神经网络分析匹配结果间的相关性问题,最后选取一个简化的军事剧情对该方法的合理性进行验证。Adaptation is the most difficult stage in the CBR cycle, especially, when the solution space is multi- dimensional in the Command Entity’s Mission Planning. This paper discusses the adaptation of a high dimensional solution space in the Command Entity’s Mission Planning and proposes a possible approach to it. A Visualization induced Self Organizing Map (ViSOM) is used to map the problem space and solution space first, then a Back Propagation (BP) network is applied to analyze the relations between these two maps. A simple military scenario is used as a case study for evaluation purposes.

关 键 词：CBR 多维解空间 适应 ViSOM 任务规划 

分 类 号：F2[经济管理—国民经济]