ABC-VMD和包络谱分析在齿轮故障诊断中的应用

针对齿轮箱故障的非线性、非稳定性特点,提出了一种参数优化变分模态分解(Variational mode decomposition,简称VMD)提取特征频率的方法。首先,利用人工蜂群算法(Artificial bee colony algorithm,简称ABC)对VMD分解的层数和惩罚因子进行自适应选择;其次,根据互信息法在VMD分解后得到的有限个本征模态函数(Intrinsic mode function,简称IMF)中选择最佳模态函数;最后,对该模态函数进行包络谱分析,有效提取齿轮故障特征频率。仿真与实验结果表明,与经验模态分解(Empirical mode decomposition,简称EMD)以及基于粒子群优化算法(Particle swarm optimization,简称PSO)的变分模态分解方法相比较,ABC-VMD方法自适应性强,可以有效克服模态混叠、信号丢失及过度分解问题,能够准确诊断齿轮箱故障,同时避免PSO-VMD易陷入局部最优的缺点。

Application of ABC-VMD and Envelope Spectrum Analysis in Gear Fault Diagnosis

Aiming at the nonlinear and unsteady characteristics of gearbox fault,a parameter-optimized Variational mode decomposition(VMD)is proposed to extract the characteristic frequency. Firstly,the Artificial bee colony algorithm(ABC)is used to adaptively select the number of layers and the penalty factor for VMD. Then,according to the mutual information method,the optimal finite Intrinsic mode function(IMF)is selected after VMD. Finally,the envelope spectrum analysis of the best IMF is performed to extract the characteristic signal of the gear fault. Comparing with Empirical mode decomposition(EMD)and VMD based on Particle swarm optimization(PSO),simulation and experimental results show that the ABC-VMD method has strong adaptability,which can effectively avoid problems of mode mixing,signal loss and excessive decomposition. It can accurately perform early fault diagnosis on the gearbox and at the same time avoid the disadvantages of PSO-VMD falling into local optimum.