本文利用粗糙集理论中的粗度概念,在对获取规则进行扩展性的粒度描述的基础上,提出了一种工具--知识依赖关系图(KDRD),构造了n维知识粒度决策空间以及基于多决策类的同构映射决策空间,建立了基于KDRD和决策空间的决策算法,并结合机械故障诊断说明了本方法非常适于故障诊断等通过结论获取原因的智能决策系统. Based on the concept of coarseness in rough set theory, this paper presents a tool - Knowledge Dependency Map (KDRD) based on the extended granularity of access rules, and constructs the n-dimensional knowledge-based decision-making space Multi-decision-making isomorphism mapping decision space, a decision-making algorithm based on KDRD and decision space is established. Combined with the mechanical fault diagnosis, this method is proved to be an intelligent decision-making system which is very suitable for fault diagnosis and other reasons.