基于HHT和FKNN的动调陀螺仪故障诊断研究  

Research on Diagnosis of Dynamically Tuned Gyroscope Based on HHT and FKNN

作　　者：吴明辉[1] 李文海[1] 李立新[2] 尚永爽[3] 

机构地区：[1]海军航空工程学院科研部 [2]空军航空大学飞行模拟训练中心 [3]94973部队

出　　处：《科学技术与工程》2014年第4期298-302,共5页Science Technology and Engineering

基　　金：总装武器装备预研基金(9140A27020212JB14311)资助

摘　　要：动调陀螺仪是一种重要的、典型的和复杂的机载机电设备。针对动调陀螺仪的故障结构特征,提出一种基于希尔伯特-黄变换(HHT)和模糊Kohonen神经网络(FKNN)的动调陀螺仪故障诊断模型。首先,在对动调陀螺仪进行故障模式与影响分析的基础上,明确了故障模式及特征参数;针对特征参数中的随机信号,采用希尔伯特-黄变换进行特征提取;然后,在分析Kohonen神经网络结构、原理及不足的基础上,提出一种模糊Kohonen神经网络诊断模型;最后,通过诊断算例进行分析验证。结果表明所提出方法在动调陀螺仪故障诊断中具有较高的正确诊断率。The dynamically tuned gyroscope is an important, typical and complex airborne electromechanical equipment. According to the fault characteristics of dynamically tuned gyroscope, a model which is based on Hil-bert-Huang Transform method and fuzzy Kohonen neural network is proposed. Firstly, based on the analysis of fail-ure mode and effects of dynamically tuned gyroscope, the failure mode and diagnosis parameters are defined. According to the random signal of the selected parameters, Hilbert-Huang Transform method is adapted to extract the feature. And then, a fuzzy Kohonen neural network diagnosis model is proposed according with analysis of structure, principle and defects of Kohonen neural network. At last, the diagnosis example shows that the proposed model has a higher accuracy in dynamically tuned gyroscope diagnosis.

关 键 词：故障诊断 动调陀螺仪 模糊Kohonen神经网络 希尔伯特-黄变换 

分 类 号：V464.3[航空宇航科学与技术—航空宇航制造工程]