摘要
针对电梯导靴振动信号采用经验模态分解(Empirical Mode Decomposition,EMD)难以直接提取早期微弱故障特征的问题,提出基于奇异值分解(Singular Value Decomposition,SVD)优化经验模态分解的电梯导靴振动信号故障特征提取方法。该方法首先对原始信号进行SVD分解,通过奇异值贡献率原则来确定相空间重组的最佳Hankel矩阵结构,利用曲率谱原则与奇异值贡献率原则相结合来确定有效奇异值的阶次;筛选出包含主要故障信息的奇异值进行信号重构,得到剔除噪声信号与光滑信号的突变信号;然后对突变信号进行EMD分解,得到信号的本征模态函数(Intrinsic Mode Function,IMF)分量。最后,对IMF分量作Hilbert变换,求得其Hilbert边际谱,从而获得电梯导靴故障特征频率信息。仿真结果表明该方法有效改善了EMD难以直接提取早期微弱故障特征的问题,更准确地提取了振动信号的故障特征频率,验证了所述方法的有效性。
The elevator guide shoe vibration signal by EMD(Empirical Mode Decomposition,EMD)to directly extract early fault features,is proposed based on the singular value decomposition(Singular Value,Decomposition,SVD)optimization of lift EMD guide shoe fault vibration signal feature extraction method.Firstly,the original signal SVD decomposition,the contribution rate of the principle to determine the optimal Hankel matrix structure reconstruction by singular value,using the principle of curvature spectrum and singular value of the contribution rate of the principle to determine the combination order of effective singular value;singular value selection mainly contains the fault information of signal reconstruction,signal mutation by removing noise signal and smooth signal;then the mutation signal is decomposed by EMD,get the signal intrinsic mode function(Intrinsic Mode,Function,IMF)components for the Hilbert transform of the IMF component to obtain the Hilbert marginal spectrum,so as to obtain the elevator guide shoe fault characteristic frequency information.The simulation results show that the method can effectively improve the problem that the EMD is difficult to directly extract the characteristic of the early weak fault,and extract the fault characteristic frequency of the vibration signal more accurately,and verify the validity of the method.
作者
兰夏燕
万舟
许有才
陶然
王家忠
和杰
杨春宇
LAN Xia-yan;WAN Zhou;XU You-cai;TAO Ran;WANG Jia-zhong;HE Jie;YANG Chun-yu(Kunming University of Science and Technology, Kunming 650500;Special Equipment Safety Inspection Institute in Yunnan Province,Kunming, 650228)
出处
《软件》
2017年第8期25-31,共7页
Software
基金
国家质检总局科技计划项目资助(2013QK104)
云南省质量技术监督局科技计划项目资助(2013ynzjkj02)