摘要
以尺度空间对信号频谱中共振频段的识别能力为基础,结合变分模态分解(VMD)对信号的自适应分解能力,提出了预估惩罚因子的尺度空间引导VMD算法。该算法的核心包括以尺度空间对信号频段的共振频段划分从而确定VMD中的本征固有模态个数,并根据共振频段边界预估VMD各个本征固有函数的初始中心频率与相应的惩罚因子取值,从而提高VMD的自适应性以及准确性。仿真结果表明,该方法能够正确识别低信噪比条件下的故障信号的共振频带,并对信号进行准确的分解。应用高速列车轴箱轴承实验数据对该方法进行实验验证,能够有效分解信号中包含的不同故障冲击;与选择不同惩罚因子的VMD算法相比,能够更准确地提取出信号中的不同故障冲击,对VMD分解的自适应性与准确性有着显著提高。
Based on the recognition ability of scale space for resonance frequency band in signal spectrum,combined with the adaptive decomposition ability of variational mode decomposition(VMD),a scale space guided VMD algorithm was proposed to predict penalty factor.The core of the algorithm included dividing resonance frequency bands of signal frequency band in scale space to determine the number of intrinsic modes in VMD,estimate the initial center frequency and corresponding penalty factor of each intrinsic function of VMD according to the boundary of resonance frequency band,and improve the adaptability and accuracy of VMD.The simulation results showed that the proposed method can recognize resonance frequency bands of a faulty signal under low SNR condition and guide VMD to correctly decompose the signal;the test data of axle box bearing of high-speed train are used to verify the proposed method being able to effectively decompose different fault shocks in signal;compared to the VMD algorithm with different penalty factors,the proposed method can more accurately extract different fault shocks,and significantly improve the adaptability and accuracy of VMD.
作者
黄衍
林建辉
刘泽潮
黄晨光
HUANG Yan;LIN Jianhui;LIU Zechao;HUANG Chenguang(State Key Lab of Traction Power,Southwest Jiaotong University,Chengdu 610000,China)
出处
《振动与冲击》
EI
CSCD
北大核心
2021年第3期240-245,共6页
Journal of Vibration and Shock
基金
国家科技部计划项目(2017YFB1201004-25)。
关键词
轴箱轴承
故障诊断
尺度空间
变分模态分解
axle box bearing
fault diagnosis
scale space
variational mode decomposition(VMD)
