摘要
实例检索是基于实例产品设计的关键,最近邻法是常用的一种检索算法。但随着实例库的增大,这种算法检索的效率会大大降低。文章结合基于CBR的自动武器设计系统的特点,提出了一种改进的算法。采用聚类的方法把实例库分为合理的聚类,并找到每个聚类的均值,然后在推理中,新实例直接与每个均值进行比较,找到与它最相近的聚类,并在这个聚类中搜索最相近的实例。避免了盲目搜索,优化了算法。
Case retrieval is one of the most important issues for research in case-based reasoning （CBR）, and the nearest neighbor algorithm is broadly adopted currently. With the growing of case library, the efficiency of nearest neighbor algorithm reduces greatly. An improving algorithm was presented according to the characteristic of automatic weapon design system based on CBR. It divided the case library into rational clustering with clustering method, and found out the mean for each duster. Then, the new case was compared directly with these means to find out the closest cluster and the nearest case. Therefore, aimless searching was avoided and the algorithm is optimized.
出处
《制造业自动化》
北大核心
2008年第11期93-95,共3页
Manufacturing Automation
基金
国防基础科研项目(K1000010601)
关键词
最近邻法
实例推理
聚类算法
自动武器设计
nearest neighbor algorithm
case-based reasoning
duster algorithm
automatic weapon design
