基于多方法联合的故障诊断技术研究  

Research on Fault Diagnosis Technology Based on Combination of Multi-methods

作　　者：刘军[1] 吴红兰[1] 宫淑丽[1] 

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1期111-115,120,共6页Aeronautical Computing Technique

基　　金：航空科学基金项目资助(2012ZC52)

摘　　要：针对基于故障树(FTA)的故障诊断方法存在的空间爆炸问题和基于神经网络的故障诊断方法存在的训练样本整理困难问题,将故障树和BAM神经网络融合的方法应用于航电系统故障诊断,利用FTA得到系统的故障模式,进而分析归纳出BAM的训练样本,通过联想记忆矩阵并行联想,得到诊断结果,扩展综合故障诊断能力。对航电系统进行故障诊断仿真分析,验证了方法的有效性。并联合基于BP神经网络故障诊断方法,将二者优点结合进行联合故障诊断,验证了联合故障诊断方法的可行性。The fusion method of fault tree and BAM neural network was applied to avionics system fault diagnosis aiming at the space explosion problem of fault diagnosis method based on fault tree（ FTA） and the difficulty in sorting the training samples in those methods based on neural network. FTA were used to get the failure mode of the system,then the training samples for BAM neural network were analyzed and summarized,finally the results of diagnosis were obtained through associative memory matrix. Thus the fusion method could expand the comprehensive fault diagnosis ability. And the effectiveness of the method had been verified by means of fault diagnosis simulation analysis on avionics system,which And joint fault diagnosis was conducted by associating with the fault diagnosis methods based on BP neural network through combining both advantages,then verified the feasibility of combining diagnosis method.

关 键 词：FTA BAM神经网络 BP神经网络 故障诊断 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]