基于PSO-VMD的齿轮特征参数提取方法研究

发布时间：2018-04-17 13:26

本文选题：齿轮 + 特征提取　；参考：《制造技术与机床》2017年11期

【摘要】：针对变分模态分解(Variational Mode Decomposition,VMD)中分解模态个数需要凭经验去预先设置这一问题,提出一种基于幅值谱的粒子群优化算法(Particle Swarm Optimization,PSO)对VMD参数进行优化的方法,并应用到齿轮故障特征参数提取中。首先,根据齿轮振动信号的特点,选取了幅值谱作为PSO的适应度函数,利用基于幅值谱的PSO求取VMD参数中的模态分解个数K和惩罚因子α;其次,使用参数组合[K,α]优化后的VMD来分解采集的齿轮数据,可得K个模态分量;最后,将K个模态分量构造出的矩阵进行奇异值分解,可得K个奇异值并将其构成特征向量,用特征向量构造欧氏距离分类器来诊断测试数据。所提出的方法应用到QPZZ-II故障仿真平台。试验结果表明,所提出的方法可以有效地提出齿轮故障的特征,得到的诊断率最高。
[Abstract]:According to the variational modal decomposition (Variational Mode Decomposition, VMD) in a number of decomposition to the problem set in advance by experience, put forward a kind of amplitude spectrum based on particle swarm optimization algorithm (Particle Swarm Optimization, PSO) the method to optimize the VMD parameters, and applied to gear fault feature extraction in the first. And according to the characteristics of gear vibration signal, the amplitude spectrum as the fitness function of PSO, using the amplitude spectrum based on PSO for the VMD parameter in the modal decomposition number K and the penalty factor; secondly, using the parameter combination of [K alpha] optimized VMD to decompose the collected data to the gear. K modal components; finally, the K modal component constructed by singular value decomposition of the matrix, we get K a singular value and the characteristic vector, a feature vector construction of Euclidean distance classifier to test the proposed diagnostic data. The method is applied to the QPZZ-II fault simulation platform. The experimental results show that the proposed method can effectively put forward the characteristics of gear fault and get the highest diagnostic rate.

【作者单位】：沈阳航空航天大学自动化学院;上海航空测控技术研究所故障诊断与健康管理技术航空科技重点实验室;
【基金】：航空科学基金(20153354005,20163354004) 国防基础科研项目(Z052012B002) 辽宁省自然科学基金(2014024003)
【分类号】：TH132.41

【相似文献】