基于CBR 的航空导弹故障智能诊断模型

针对航空导弹故障现象多，故障原因复杂，传统故障诊断方法难以准确进行故障定位的不足，构建一种基于案例推理的航空导弹智能故障诊断模型。结合维修实践中积累的故障案例，运用层次分类法与改进的聚类分析法，设计三级案例库组织结构，提出分阶段的K-近邻检索策略，将粗糙集理论应用于解决故障案例的征兆权重分配问题，并开发基于We b浏览形式的航空导弹智能故障诊断模块。该故障诊断模型具有较高的检索准确率，智能化程度高，使用便捷，能有效辅助技术人员进行航空导弹的故障定位和维修排故工作。

Design of Aviation Missile Fault Intelligent Diagnosis Model Based on CBR

In accordance with the defect that aviation missile has many faults, and the fault reason is complicated, and the traditional fault diagnosis method is difficult to find out fault accurately, the fault diagnosis model is established. The thesis combines organically classification with improved cluster analysis, designing three-level-structure for cases and classifying accurately fault cases to reduce greatly the scope of cases search. The phase-divided K-nearest neighbor (K-NN) strategy is advanced to eliminate the inefficiency of cases search for using directly K-nearest neighbor in many cases. In order to overcome subjectivity of expert evaluation, rough sets theory is used to resolve the omen weight distribution of fault cases so that the weight has objectivity, and it can ensure the precise of cases search. The integrated fault diagnosis system based on Web is developed by the above-mentioned model. The actual use indicates that the model has higher retrieval precision, high intelligent degree, easy to use, it can effectively help the maintainer positioning and removing fault.