中文摘要：

成本决策是成本管理的重要环节,为了将先进的人工智能方法融入生产成本决策过程中,并使得决策因素最小化,在深入研究粗糙集理论的基础上,提出了一种新的基于区分矩阵的最小属性约简方法,并详细阐述了该方法的含义和性质,以及算法的搜索策略与执行过程,有效解决了任意相容决策表的最小属性约简问题,并结合现有改进的值约简算法,达到了决策表规则生成最简化的目标。最后通过对某钢铁企业的实际应用,验证了方法的可行性和实用性。

英文摘要：

Cost decision is an important part in the cost management.In order to integrate the methods of advanced artificial intelligence into the process of production cost decision,and minimize the factors of decision making,on the basis of studying rough sets theory,an original method about minimal attribute reduction based on discernibility matrix is proposed by this paper.And its concepts and properties,as well as algorithm＇s search strategies and execution flow are explained in detail,which can effectively solve the problem about minimal attribute reduction in a given consistent decision table.Furthermore,combining an improved algorithm about value reduction,the most simplified rules of decision table can be generated.Finally,through applying these methods to exploit the decision table of certain iron and steel enterprises,the results have demonstrated the feasibility and practicability of the original method about minimal attribute reduction based on discernibility matrix and the improved algorithm about value reduction.