针对变分模态分解(Variational Mode Decomposition,VMD)过程中模态分量个数和惩罚参数大小依赖先验知识,单一或顺序优化单一参数可能导致局部最优的问题,提出以包络熵和包络峭度因子作为适应度函数,利用遗传算法全局寻优的特点,对VMD的模态分量个数和惩罚参数组合进行优化.通过最优参数组合下的VMD对信号进行分解,可以获得多个本征模态分量(Intrinsic Mode Function,IMF),选择适应度函数最小IMF分量作为有效IMF分量进行包络解调,从中提取轴承信号的故障特征频率.对多种轴承故障类型信号进行分析并与其他方法对比,结果表明所提方法能有效提取轴承故障特征,有助于实现微弱故障条件下轴承故障特征频率的准确提取.
In the process of Variational Mode Decomposition(VMD),the number of modal components and the size of penalty parameters depend on prior knowledge,so single or sequential optimization of a single parameter may lead to local optimality.In this paper,taking envelope entropy and envelope kurtosis factor as fitness functions,and using the characteristics of global optimization of genetic algorithm,the number of modal components and the combination of penalty parameters of VMD are optimized.Multiple Intrinsic Mode functions(IMF)can be obtained by decompressing the signal through VMD under the optimal parameter combination.The IMF component with the smallest fitness function is selected as the effective IMF component for envelope demodulation,from which fault characteristic frequencies of bearing signals are extracted.The analysis of various bearing fault signals and the comparison with other methods show that the proposed method can effectively extract bearing fault features,which is helpful to accurately extract bearing fault feature frequencies under weak fault conditions.