基于改进LMD与BP神经网络的变速箱故障诊断  

Gearbox Fault Diagnosis based on Improved LMD and BP Neural Network

作　　者：何雷 刘溯奇[2] 蒋婷[1] 黄志杰[1] He Lei;Liu Suqi;Jiang Ting;Huang Zhijie(School of Dynamic Technology,Liuzhou Railway Vocational and Technical College,Liuzhou 545000,China;College of Mechanical and Electronic Engineering,Central South University,Changsha 410083,China)

机构地区：[1]柳州铁道职业技术学院动力技术学院,广西柳州545000 [2]中南大学机电工程学院,湖南长沙410083

出　　处：《机械传动》2020年第1期171-176,共6页Journal of Mechanical Transmission

基　　金：广西高校中青年教师科研基础项目(2019KY1553);教育部支撑技术项目(625010339)

摘　　要：针对军用装甲车变速箱工作环境恶劣、故障模式难以识别的问题,在现有方法基础上,将噪声辅助分析方法、局部均值分解(LMD)方法和BP神经网络方法相结合,应用于装甲车变速箱故障诊断中。首先,在自行搭建的实验台上采集变速箱正常、轴承间隙故障、外环压痕、齿轮断齿4种典型状态下的振动信号;然后,采用噪声辅助LMD方法对信号进行分解,将信号前8个PF分量进行能量特征值提取,将提取的特征值作为BP神经网络的输入量,根据输出结果识别变速箱的故障类型。结果表明,该方法能有效应用于军用装甲车变速箱故障诊断,诊断正确率达到92. 5%。研究为其他特种变速箱诊断提供了一种有效的参考途径,有一定工程实用价值。Aiming at the problem of the poor working environment and the fault mode is difficult to identi-fy of the gearbox of military armored vehicles,based on the existing methods,the noise assisted analysis,LMDand BP neural network are combined to apply to the fault diagnosis of armored vehicle gearbox. Firstly,the vi-bration signals under the four typical states of normal gearbox,bearing clearance fault,indentation of outer ringand broken tooth of gear of the gearbox are collected on the self-built test bench. Then,the signal is decom-posed by the noise-assisted LMD method,and the energy eigenvalues of the first eight PF components are ex-tracted. The extracted feature values are used as the input of the BP neural network. The fault type of the gear-box is identified based on the output result. The results show that the method can be effectively applied to thefault diagnosis of military armored vehicle gearbox,and the diagnostic accuracy rate is 92.5%. An effective ref-erence way for other special gearbox diagnosis is provided by this study,and it has certain engineering practicalvalue.

关 键 词：变速箱 局部均值分解(LMD) 噪声辅助 BP神经网络 故障诊断 

分 类 号：E923.1[军事—军事装备学]