摘要
为增强故障诊断专家系统处理不完备信息和自动获取规则知识的能力,应用粗集理论及动态约简思想,深入研究并初步建立了故障规则自动获取系统。其中关键技术包括:故障样本信息的关系型决策表表示;决策表子表的随机采样;条件属性的动态约简算法设计;约简集合的过滤;故障规则的表达结构定义;故障规则的提取和规则的评价等。开发了相应的基于VC++平台的计算机软件,并以汽车发动机异响类故障为样本,对系统的故障规则获取过程进行了案例分析,从相关的实验结果和数据来看,较好地实现了系统的预定目标和功能。
The paper presents a fault-rule acquisition system by applying the rough set theory and the dynamic reduction technique. The key technologies adopted in the system include the way of representing the fault-sample as a relation decision table, the random sampling of sub-tables of the decision table, the design of the dynamic reduction algorithm of the condition attributes, the filtration of the reduction set, the definition of the fault-rule structure, and the acquisition and the evaluation of the final fault-rules. A relevant software is developed with VC. As an example,the procedure of the automatic acquisition of the fault-rules of an unusual automobile engine noise-fault is presented and analyzed,the result of which proves to be effective.
出处
《浙江工业大学学报》
CAS
2004年第2期203-207,共5页
Journal of Zhejiang University of Technology
