基于MNN的主动力装置故障诊断研究  

Research on Fault Diagnosis of Main Power System Based on MNN

作　　者：杨广[1] 梁晶晶[1] 

机构地区：[1]中国人民解放军91918部队

出　　处：《盐城工学院学报（自然科学版）》2014年第3期1-5,共5页Journal of Yancheng Institute of Technology:Natural Science Edition

基　　金：国家自然科学基金资助项目(60774029)

摘　　要：单神经网络诊断大多已成功应用到实际的故障诊断系统中。而在实际应用中,多层次多故障的复杂系统普遍存在,运用单一结构神经网络在处理复杂故障诊断问题时网络过于庞大,诊断精度降低。对层次分类诊断模型进行研究分析,提出运用多重结构神经网络(MNN)来对主动力装置进行故障诊断研究,诊断结果表明该模型是可行和有效的。Artificial neural network （ANN） has been successfully applied to fault diagnosis systems in real -world applications; however, what is mostly used for diagnosing faults is single network. Factually multiple faults occur frequently at different levels in complicated systems and when the single network is used to deal with complicated problems of fault diagnosis, it＇ 11 be so gigantic that a series of difficulties will be brought to network training. In this paper, on the basis of analysis of hierarchical classified diagnostic model, MNN is presented to research on fault diagnosis of main power equipment and the diagnostic result indicates that the model is feasible and valid.

关 键 词：主动力装置 故障诊断 神经网络 MNN 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置] O212.6[自动化与计算机技术—控制科学与工程]