摘要
以VMD-DE方法对行星齿轮箱进行了故障诊断测试,当VMD分解层数确定后,再通过VMD分解振动信号获得IMF分量,之后计算IMF和原信号归一化互信息值,通过筛选得到归一化互信息比定阈值更大的IMF并完成信号重构,之后计算得到重构信号DE特征值,再输入PSO-SVM中并识别行星齿轮故障模式。研究结果表明:各状态下的DE值都发生了大幅波动,正常状态与z15剥落故障下形成的DE发生了相互交叉。采用粒子群优化方法计算出SVM惩罚因子为C=0.45,高斯核参数为σ=25.13,达到了稳定的适应度。采用VMD信号分解的方法来实现信号的重构降噪过程,由此得到更优分类结果,由此可以推断以VMD重构后能够显著降低信号噪声。三种状态都实现了正确分类,达到了100%的准确率。
The method of VMD-DE is applied to the fault diagnosis of a planetary gear box.After the VMD decomposition layers is determined,the IMF component is obtained by decomposing the vibration signal.The normalized mutual information of the IMF and the original signal is subsequently obtained by screening signals above the set threshold.The signal reconstruction is completed and the DE characteristics of the reconstructed signal are obtained,which are then input into the PSO-SVM to determine planetary gear fault modes.The results show that the DE values fluctuate greatly in each state,and the DE values formed in the normal state and the Z15 spalling fault cross each other.The SVM penalty factor C=0.45 and the Gaussian kernel parameterσ=25.13 were determined by particle swarm optimization method,which reached stability.The VMD signal decomposition method is used to realize signal reconstruction and noise reduction to obtain better classification results.It can be inferred that the signal noise can be significantly reduced after VMD reconstruction.Correct classification was achieved in all three states,with 100%accuracy.
作者
吴秋梅
李强
于艳
WU Qiumei;LI Qiang;YU Yan(Intelligent Manufacturing College,Xinxiang Vocational and Technical College,Xinxiang Henan 453000,China;School of Mechanical Engineering,Henan Polytechnic University,Zhengzhou 450000,China;Aviation Industry Xinhang Yubei Steering System(Xinxiang)Co.,Ltd.,Xinxiang Henan 453000,China)
出处
《机械设计与研究》
CSCD
北大核心
2022年第2期101-104,共4页
Machine Design And Research
基金
河南省高等学校重点科研项目计划资助(18A460002)。
关键词
行星变速箱
故障诊断
变分模态分解
散布熵
支持向量机
准确率
planetary gearbox
fault diagnosis
variational modal decomposition
spread the entropy
support vector machine
accuracy
