一种改进的基于差别矩阵属性约简算法及其应用  

An Improved Attribute Reduction Algorithm Based on Discernibility Matrix and Its Application

作　　者：纪赖恩[1] 李秀娟[1] 许晓东[1] 王湘予[1] 

机构地区：[1]河南工业大学电气工程学院,郑州450007

出　　处：《计算机与数字工程》2010年第6期13-16,共4页Computer & Digital Engineering

基　　金："十一五"国家支撑计划项目(编号:2008BADA8B0402);河南省科技攻关项目(编号:092102210008)资助

摘　　要：为得到决策表的一个最优约简或相对最优约简,提出了一种改进的约简算法,先利用差别矩阵约简算法得出约简全集,然后通过定义约简的可信度、相对于核的属性重要度,计算出决策表的约简,并在中央空调系统节能决策表中应用,结果表明了该算法的有效性。For obtaining the optimal or a relative optimal reduction of the decision-making table,an improved reduction algorithm is presented in the paper.Firstly,the whole reduction set is calculated by the reduction algorithm based on descernibility matrix.Secondly,the reduction decision table is computed by the definition of the credibility of reduction and the importance of the attributes which relative to the core,and it is used in the decision table of the energy saving of central air-conditioning system.The experiment illustrates the algorithm is effective.

关 键 词：差别矩阵 约简的可信度 属性重要度 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]