CAJ | 학술논문

针对某型导弹故障测试数据的部分缺失问题，提出一种基于双概率相似关系的不完备决策系统数据补齐算法。在ROUSTIDA算法的基础上，根据决策属性值将不完备决策系统划分为各个决策子系统，考虑不同决策子系统中相似对象对决策规则的影响，选择同一决策子系统内最相似的对象来填补条件属性缺失值，并通过实例进行验证分析。结果表明：该算法在一定程度上能降低计算的复杂性、提高填补率，并能有效解决ROUSTIDA算法可能存在的决策规则冲突问题。
침대모형도탄고장측시수거적부분결실문제，제출일충기우쌍개솔상사관계적불완비결책계통수거보제산법。재ROUSTIDA산법적기출상，근거결책속성치장불완비결책계통화분위각개결책자계통，고필불동결책자계통중상사대상대결책규칙적영향，선택동일결책자계통내최상사적대상래전보조건속성결실치，병통과실례진행험증분석。결과표명：해산법재일정정도상능강저계산적복잡성、제고전보솔，병능유효해결ROUSTIDA산법가능존재적결책규칙충돌문제。
According to the problem of the lack of fault test data of a certain type of missile, the paper puts forward a completing data algorithm based on double probability similarity in incomplete decision system. In accordance with the decision attribute values, the incomplete decision system was divided into various decision subsystem by the algorithm based on ROUSTIDA algorithm. Considering that similar objects in different decision subsystem may affect the decision-making, the algorithm of this paper completed the losing value of condition attributes with the most similar objects in the same decision subsystem, which was analyzed with example. The results prove that the algorithm can partly reduce the computational complexity, improve the completing rate and effectively solve the problem that decision-making principle may exist conflict by ROUSTIDA algorithm.