【作者】 王文熙；

【导师】 刘青宝；

【作者基本信息】 国防科学技术大学 ， 管理科学与工程， 2010， 硕士

【摘要】 随着新军事变革的不断深入和发展,信息技术所带来的数据容量急剧增长,以数据挖掘等信息处理技术为动力的军事转型也势在必行。数据挖掘是未来信息化战争中掌握信息化优势,牢牢掌握战争主动权的得力工具。装备保障体系中,装备维修器材是实施装备维修保障的重要物质基础,对装备的战备完好性和战斗力具有重要的影响。随着装备复杂度的提高,维修器材的确定与优化问题也越来越突出。发现维修器材之间消耗的规律对维修器材保障的优化决策具有重要的意义,能够提高保障的效率。关联规则挖掘是数据挖掘中最重要的任务之一,其目标是发现数据库中属性之间的关联关系,为维修器材保障的优化提供了一种有效的解决方法。本文在研究数据挖掘中关联规则挖掘算法的基础上,结合问题的特点,着重研究了关联规则的一种扩展形式——模糊关联规则。针对现有的模糊关联规则挖掘算法的不足之处,借鉴关联规则经典算法Apriori算法和FP-tree算法的特点,提出了一种基于线性链表的模糊关联规则挖掘算法。该算法只对事务数据库进行一次扫描,仅记录对计算频繁项集支持度有贡献的事务信息,减少了数据存储的开销,提高了算法计算效率。通过时间复杂度分析证明了算法效率的高效性。通过UCI数据集对算法的准确性和有效性进行了实验验证。最后,将算法应用于对装备维修器材保障优化问题的解决。本文首次采用模糊关联规则的挖掘方法对装备维修器材保障优化问题进行研究,提出了一种效果良好、性能优越的模糊关联规则挖掘算法,并采用CRISP-DM数据挖掘方法对挖掘过程进行了建模分析,为我军装备维修器材保障的优化提供了一种有价值的技术参考。更多还原

【Abstract】 With the intensification and development of the new military innovation, information technologies arouse the rapid increase of data, military transform has to be happened via information processing technology such as data mining.Data mining is powerful tool for getting an informationize advantage and seizing the initiative in the informationize war in the future. At the arming support system,the arming maintenance materials are the important elements of the arming maintenance support,have an effect on the integrality and performance of arming. With the increasing complexity of arming, the choice and optimizing of maintenance materials become more and more important. Discovering the rules of maintenance materials using is signality for the optimizing decision-making,and can improve the efficiency of arming maintenance materials support. As one of the main tasks in the field of data mining,association rule mining is used to discover the relationships among the attributes,and it is a useful approach to address the problem of optimizing maintenance materials support.Based on the research of association rule mining algorithm and considering the feature of the problem, this thesis puts the research attention to one of the association extend forms——fuzzy association rule. In order to improve the efficiency of existing fuzzy association rule mining algorithm, we proposed a mining algorithm base on linear linklist in this thesis via learning the merits of the classic association rule mining algorithm Apriori and FP-tree. This algorithm scans database once, and only records the information of the transactions, which are useful for counting the support of the frequent itemset. It reduces the spending of data storage, increase the running efficiency. We prove this algorithm has high performance via the time complexity analysis. We also have tested the veracity and validity of this algorithm by experiment on a dataset from UCI. And last,we use this algorithm to solve the optimizing problem of maintenance materials support.In this thesis, we first use fuzzy association rule minimg method to solve the optimizing problem of maintenance materials support, propose a excellent mining algorithm, and build the mining modeling via using a data mining methodology called CRISP-DM. It is a valuable reference for arming maintenance materials support for our army.更多还原