【摘要】 以构建发动机故障自动诊断系统为出发点 ,指出缺乏知识是技术实现的难点所在。以全局的观点分析了发动机的知识范围 ,采用粗糙集理论从数据中挖掘规则知识 ,给出了规则可信度的确定方法 ,并采用粗糙集理论中的属性重要性概念来对CBR理论中实例相似度权重进行分配 ,避免了主观性。考虑到发动机故障关联性强、故障现象和原因之间少有一一对应的特点 ,提出了基于结构分解的知识组织模型 ,并集成多种方法来共同完成故障诊断更多还原

【Abstract】 Aiming to construct automatic diagnostic system for engine faults, we pointed out the technical bottleneck is the lack of knowledge, and based on the analysis of the whole knowledge scope of engines, a methodology applying rough sets to mining rules from data was brought out, which can and compute the confidence coefficient of rules, and the method to distribute the weightiness of comparability in CBR(case-based reasoning) was given. Considering the characteristics of engine faults were interrelated and multi-mapped between symbols and causes, so a knowledge organization tree based on the engine structure is described and the nodes that are knowledge carriers are modeled. Lastly, an integrated method to fulfill fault diagnosis is implemented.更多还原