摘要
为从液压系统振动信号中提取有效特征进行故障诊断,针对随机噪声、端点效应和虚假分量会影响变分模态分解(VMD)的分解精度问题,提出了一种改进VMD的故障特征提取方法。首先,针对随机噪声会导致分解误差增大现象,提出了基于奇异值差分谱降噪预处理,该方法能抑制噪声对分解结果的干扰;然后,针对端点效应会导致VMD处理信号两端产生明显的飞翼现象,提出了基于支持向量回归机的端点延拓,该方法具有较高的拟合精度;最后,针对虚假本征模态函数(IMF)分量会导致VMD处理出现能量泄漏现象,提出了IMF能量熵增量的虚假分量剔除,该方法的真假分量具有区分性。仿真信号和实测液压信号分析表明:改进VMD能有效改善传统VMD方法在特征提取上的三个不足,可准确提取液压故障信号的主要特征频率,实现液压系统故障的精确诊断。
In order to extract effective features from the vibration signal of hydraulic system for fault diagnosis,an improved VMD method was proposed to solve the problem that random noise,endpoint effect and false component would affect the resolution accuracy of variational mode decomposition(VMD).Firstly,in view of the influence of random noise on the decomposition accuracy,a pre-processing approach of reducing the random noise was researched based on singular value decomposition(SVD)difference spectrum,which can suppress the noise interference to the decomposition accuracy.Then,in view of the fact that the endpoint effect will lead to the obvious wing phenomenon at both ends of VMD decomposition,the endpoint extension method based on support vector regression machine(SVRM)was proposed,which had a higher fitting accuracy.Finally,in view of the fact that the false intrinsic mode function(IMF)component will lead to the energy leakage phenomenon in VMD decomposition,the elimination of the false IMF component based on the energy entropy increment,and the difference between the true and false components was obvious.The analysis of simulation signal and measured hydraulic signal show that the improved VMD can effectively improve the three shortcomings of VMD method in feature extraction,and accurately extract the main characteristic frequency of hydraulic fault signal,which can realize the fault accurate diagnosis of hydraulic system.
作者
丰少伟
柴凯
朱石坚
杨庆超
楼京俊
FENG Shao-wei;CHAI Kai;ZHU Shi-jian;YANG Qing-chao;LOU Jing-jun(College of Naval Architecture and Ocean Engineering, Naval Univ. of Engineering, Wuhan 430033, China)
出处
《海军工程大学学报》
CAS
北大核心
2021年第2期6-13,29,共9页
Journal of Naval University of Engineering
基金
国家自然科学基金资助项目(51679245,51579242,51509253)
湖北省自然科学基金资助项目(2018CFB182)。
关键词
液压系统
变分模态分解
奇异值差分谱
支持向量回归机
本征模态函数
能量熵增量
hydraulic system
variational mode decomposition
singular value decomposition diffe-rence spectrum
support vector regression machine
intrinsic mode function
energy entropy increment
