基于多方法联合的故障诊断专家系统设计  

Design and Validation of Avionics Fault Diagnosis Expert System Based on Combination of Multi- methods

作　　者：姬志伟[1] 刘军[1] 孙益祥[1] 符积宏 

机构地区：[1]南京航空航天大学民航学院,江苏南京211106

出　　处：《航空计算技术》2015年第5期93-96,共4页Aeronautical Computing Technique

基　　金：国家自然科学基金与民航总局基金联合资助项目(MHRD201125)

摘　　要：针对传统基于单一方法的航电故障诊断专家系统存在的数据爆炸、知识获取代价与扩展难度大、诊断效率低等问题,提出了一种基于FTA、BAM神经网络、BP神经网络方法联合的故障诊断专家系统模型,模型保留了FTA对各种故障事件优秀的分析处理能力,发挥了BAM神经网络对单一或多源故障的快速定位及BP神经网络处理电路故障的优势,三者结合能够准确地找出故障位置和原因。利用VS2010平台研发了某型飞机航电故障诊断专家系统,以航电系统典型故障为诊断实例,验证了该专家系统诊断速度快、诊断结果准确,解决了传统专家系统存在的问题。In the view of the traditional avionics expert diagnosis system based on single method,which exist problems such as data explosion,knowledge acquisition cost and difficult extension and low diagnosis efficiency,it puts forward a model of expert diagnosis system which based on the FTA,BAM and BP neural network.This model retains the excellent ability of FTA which can deal with vorious failure event.It develop the advantages of BAM and BP neural network which can rapid locate the single and multiple source fault and process circuit malfunction.It can find out the fault position and cause accurately when comebine them together .Using VS2010 platform develops a certain type of aircraft avionic fault diagnosis expert system.Taking the typical fault of avionics system as a diagnosis example,which verify the expert system .It can diagnoe fast and accurate,which solves the problems existing in the traditional expert system.

关 键 词：故障诊断 专家系统 FTA BAM神经网络 BP神经网络 航电系统 

分 类 号：V240.2[航空宇航科学与技术—飞行器设计]