摘要
针对空空导弹仿真系统中传统数据分析方法效率低下,无法快速有效对大量仿真数据进行故障识别的问题,提出基于长短时记忆网络的仿真系统数据故障诊断方法。该方法分析仿真试验数据的数据特征,采用长短时记忆网络对健康数据中的时序信息进行记忆,并内嵌至所建立的多层次深度神经网络模型,从而有效表征健康数据的本征变化属性,进而对试验过程中的故障数据进行准确诊断。空空导弹仿真系统数据智能诊断结果表明,本方法在快速诊断的基础上,故障诊断识别率达到90%以上。
Aiming at the problem that the traditional data analysis methods in air to air missile simulation system is inefficient,and can not quickly identify faults in a large number of simulation data,a data fault diagnosis method based on long and short term memory network was proposed.The method analyzed the data characteristics of the simulated experimental data,used the long and short term memory network to memorize the time series information in the health data,and embedded it into the established multi-level deep neural network model,thereby to effectively characterized the intrinsic change attribute of the health data.In turn,an accurate diagnosis of the fault data during the test was performed.The data intelligent diagnosis results of air to air missile simulation system showed that the recognition rate of fault diagnosis was over 90%.
作者
牛群
刘志永
褚建川
王艳奎
吴根水
NIU Qun;LIU Zhiyong;CHU Jianchuan;WANG Yankui;WU Genshui(China Air Missile Institude,Luoyang 471099,China)
出处
《探测与控制学报》
CSCD
北大核心
2019年第5期25-29,共5页
Journal of Detection & Control
关键词
故障诊断
长短时网络
神经网络
数据分析
failure diagnosis
long short-term memory
neural network
data analysis
